Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 431 appraisals
Journal of medical Internet research
Advancing Human-Centered AI in Clinical Decision Support: Sociocognitive Human-in-the-Loop Study in HIV Care
BACKGROUND: AI-powered clinical decision support systems (CDSS) have shown promise in improving prediction, monitoring, and treatment optimization across clinical domains, including HIV care. However, translating AI outputs derived from electronic health records into clinically meaningful, trustworthy, and actionable decision support remains challenging, underscoring the need for more human-centered and socioecologically grounded CDSS design. OBJECTIVE: This study aimed to explore how we can effectively translate the outputs of machine learning models based on HIV electronic health records into a real AI-powered CDSS for HIV care. Using the human-in-the-loop method, we engaged a set of stakeholders, including HIV physicians, nurse practitioners, infectious disease pharmacists, social workers, and case managers. Stakeholders interacted with an AI-powered CDSS prototype to identify barriers and challenges to adoption, as well as to inform a more holistic and context-aware AI-powered CDSS design. METHODS: We conducted a field study at Prisma Health in South Carolina that included pre- and postsurveys, interactive usability testing sessions, think-alouds, and in-depth interviews with 16 clinicians providing HIV care between March and September 2025. We analyzed survey responses using descriptive statistics, and then transcribed and analyzed think-aloud and interview data using an etic and emic approach. RESULTS: Clinicians identified multiple challenges and design considerations for AI-powered HIV CDSS, demonstrating that clinician-AI interaction is inherently sociotechnical and embedded across multiple socioecological levels. While clinicians relied on familiar clinical indicators as cognitive anchors for interpreting AI predictions, they emphasized that social determinants of health were central to their own risk assessment and clinical decision-making. Additionally, clinicians' trust in AI is conditional and develops over time, with explainability and actionability emerging as critical factors for translating predictions into meaningful clinical interventions. CONCLUSIONS: Findings highlight the need to move beyond technically accurate predictions toward AI-powered CDSS designs that align with clinicians' cognitive practices and socioecological realities of HIV care. By extending a sociocognitive framework through empirical grounding in HIV clinical practice, this study offers design insights for developing AI-powered CDSS that are trustworthy, context-aware, and capable of supporting actionable decision-making in HIV care settings and beyond.
2 Aug 2026
Read appraisal →Neurological research
Federated deep learning model for epilepsy seizure detection using electroencephalogram (EEG) signal
OBJECTIVES: Epilepsy is a chronic neurological disorder characterized by recurrent seizures due to abnormal brain activity, which affects individuals' health and quality of life. Traditional seizure detection methods face challenges related to data privacy and security as well as difficulty in fully capturing both temporal and spatial relationships within the Electroencephalography signal. To address these limitations, a Federated Learning Enabled Unified Transformer model is proposed. METHODS: The Federated Learning with Paillier Homomorphic Encryption is deployed for preserving data privacy and enabling collaborative model training. Adaptive noise filtering and Independent Component Analysis are deployed to remove the noise and artifacts. The Multi-Scale Wavelet Coefficient is applied for signal decomposition that effectively extracts seizure-related features by decomposing Electroencephalography signals into multiple sub-bands. The Hybrid Graph-Based Attention Framework integrated Edge-Enhanced Graph Convolutional Networks for spatial feature extraction and Spectral Graph Attention for frequency-based feature selection, and refining feature representation for improving the classification accuracy. Further, the proposed technique utilizes a Unified Transformer model for seizure classification that efficiently captures temporal and spatial dependencies in Electroencephalography signals. RESULTS: The proposed model is validated on three datasets and attains 98.91% accuracy, 98.93% security, and 98.82% precision. The simulation outcomes indicate that the Federated Learning Enabled Unified Transformer model achieved outstanding performance when compared to existing models. DISCUSSION: The Federated Learning Enabled Unified Transformer model provided a superior performance by integrating Federated learning and hybrid Deep Learning models. It ensures that the model is more suitable for healthcare users.
2 Aug 2026
Read appraisal →Neural networks : the official journal of the International Neural Network Society
MFBD: Model-free backdoor defense based on vision-language pre-trained models
Backdoor attacks pose serious security threats to deep neural networks (DNNs), causing networks to output predetermined target labels when specific trigger patterns are embedded in input images. While existing backdoor defense methods utilize additional clean data or victim network models, their effectiveness in defending against various trigger injection methods is limited. We find that vision-language models, such as BLIP-2 and CLIP, are largely unaffected by backdoor triggers in poisoned images. Based on this observation, we propose MFBD, an effective model-free backdoor defense method leveraging the semantic consistency of dual vision-language descriptors. Specifically, we employ BLIP-2 and CLIP to generate descriptions of an image and its saliency masked image, respectively. We then embed both the descriptions and the image labels into a semantic space using SBERT. To distinguish clean and poisoned images, we compute the cosine similarity between their semantic representations. The image exhibiting a low cosine similarity value is classified as a poisoned image. We evaluate a suiute of backdoor defense methods against various attack methods on three datasets. Experimental results demonstrate that our method can effectively mitigate various complex backdoor threats while maintaining high performance across different datasets.
2 Aug 2026
Read appraisal →Neural networks : the official journal of the International Neural Network Society
An approximation theory perspective on machine learning
A central problem in machine learning is often formulated as follows: Given a dataset [Formula: see text] , which is a sample drawn from an unknown probability distribution, the goal is to construct a functional model f such that f(x) ≈ y for any (x, y) drawn from the same distribution. Neural networks and kernel-based methods are commonly employed for this task due to their capacity for fast and parallel computation. The approximation capabilities, or expressive power, of these methods have been extensively studied over the past 35 years. In this paper, we will present examples of key ideas in this area found in the literature. We will discuss emerging trends in machine learning including the role of shallow/deep networks, approximation on manifolds, physics-informed neural surrogates, neural operators, and transformer architectures. Despite function approximation being a fundamental problem in machine learning, approximation theory does not play a central role in the theoretical foundations of the field. One unfortunate consequence of this disconnect is that it is often unclear how well trained models will generalize to unseen or unlabeled data. In this review, we examine some of the shortcomings of the current machine learning framework and explore the reasons for the gap between approximation theory and machine learning practice. We will then review some of recent work that achieves function approximation on unknown manifolds without the need to learn specific manifold features, such as the eigen-decomposition of the Laplace-Beltrami operator or atlas construction. In many machine learning problems, particularly classification tasks, the labels yj are drawn from a finite set of values. We summarize another recent paper that establishes a deep connection between signal separation problems and classification problems, proposing that classification tasks should be approached as instances of signal separation. We conclude by identifying several open research problems that warrant further investigation.
2 Aug 2026
Read appraisal →Current opinion in critical care
Smart feeding: the role of artificial intelligence and integrated nutrition platforms in the ICU
PURPOSE OF REVIEW: Tremendous improvement in the use of artificial intelligence has opened new opportunities to analyze the data obtained from electronic health records and imaging. New technologies have tried to overcome obstacles to implement guidelines and recommendations. This review aims to describe the recent progress in the use of machine learning and new technologies in the field of nutrition of the critically ill. RECENT FINDINGS: Increase in data availability, ability to extract these data and analyze them using machine learning has allowed data scientists together with ICU specialists to improve nutritional screening and assessment and to predict occurrence of obstacles like enteral feeding intolerance or refeeding hypophosphatemia. In addition, new technologies can ensure nasogastric tube positioning and enteral feeding efficacy. Integrated platforms can integrate nutritional needs with most adequate prescriptions and modulate the nutritional administration according to the patient's tolerance and requirements. Analysis of continuous recording of imaging obtained from ultrasound can also predict gastric intolerance. SUMMARY: Using machine learning, numerous algorithms and nomograms have been suggested to predict enteral feeding intolerance but validation of these predictions is still required. New technologies integrating energy requirements and delivery of the optimal enteral feeding are very promising.
2 Aug 2026
Read appraisal →Seminars in musculoskeletal radiology
Artificial Intelligence (Pattern Recognition) in Musculoskeletal Imaging: The Future or Hype?
Artificial intelligence and automated pattern recognition, in particular, have been described as the next frontier in musculoskeletal imaging. However, as the initial hype phase transitions into clinical reality, an essential evaluation of these technologies is required. This narrative review examines the dichotomy between a potential future in which artificial intelligence offers unprecedented efficiency in automatizing multiple tasks in musculoskeletal radiology, from fracture detection, automated segmentation, to automated reporting, versus the hype, characterized by deep learning models that lack generalizability across different scanner vendors and patient populations. We explore the black box nature of deep learning and the ethical implications of automation. By analyzing current barriers to deployment, including workflow integration and regulatory hurdles, this article argues that although artificial intelligence holds transformative potential for musculoskeletal radiology, its success depends on moving beyond narrow diagnostic tasks toward robust multi-institutional validation, to evolve from a speculative trend into an essential clinical copilot tool.
2 Aug 2026
Read appraisal →Neural networks : the official journal of the International Neural Network Society
Deep learning algorithms for license plate recognition: A review.
License plate recognition serves as a crucial link in the intelligent transportation system and vehicle management. It lies at the heart of enhancing road supervision efficiency and enabling automated services. However, the rapid advancement in this field has resulted in numerous parallel technological pathways, disparate performance assessment standards, and ambiguous future trends. Consequently, there is a pressing need to clarify the technological evolution, evaluate the efficacy and limitations of various methods in complex scenarios, and thereby identify future research directions. To address this gap, we present a systematic review and analysis of the field. Deep learning technology applied in this domain overcomes the limitations of traditional image-processing methods in complex scenarios. This application notably boosts the robustness of license plate localization and character recognition, attracting extensive attention from academia and industry in recent years. In this paper, we first collate and present the public mainstream license plate image datasets and their characteristics. Subsequently, we systematically summarize the evolution of license plate detection and recognition technology. It ranges from traditional methods like edge detection and morphological segmentation based on feature engineering to end-to-end detection frameworks founded on deep learning and character recognition models. We compare and analyze the innovation aspects and application limitations of each algorithm. Through in-depth exploration of the field literature, we identify that the main challenges remain illumination intensity variation, extreme weather interference, multi-angle tilting, non-standard license plate designs, low-resolution images, and character adhesion. The research further indicates that constructing lightweight networks integrating multi-scale features, developing cross-regional license plate universal models, and incorporating traffic-scene prior knowledge into optimization algorithms will be important development directions. Meanwhile, cross-modal methods such as video-stream timing analysis and multi-sensor data fusion are being explored, which are expected to enhance real-time recognition performance in dynamic scenes. This paper offers a theoretical reference for the optimization and innovation of license plate recognition technology in the intelligent transportation system.
1 Aug 2026
Read appraisal →Artificial intelligence in medicine
Green artificial intelligence in health applications
As the healthcare sector increasingly integrates Artificial Intelligence (AI) technologies to improve operational effectiveness, diagnosis, and therapy, the environmental footprint of these innovations has become a growing concern. High energy consumption, electronic waste, and carbon emissions associated with the deployment and training of AI models pose sustainability challenges that must be addressed. This paper investigates the concept and application of Green AI in healthcare, aiming to balance technological advancement with environmental responsibility. This systematic review explores key themes, including green computing practices, the adoption of energy-efficient AI models, and the use of renewable energy sources within healthcare settings. It identifies a range of healthcare applications employing Green AI, highlights emerging trends, and emphasizes the growing importance of environmental awareness in AI development. Furthermore, the study examines enabling tools and techniques, outlines barriers to adoption, and highlights how Green AI can help streamline processes, reduce resource waste, and promote environmentally friendly medical procedures like telemedicine. The review also discusses the significance of policy frameworks, international initiatives, and cross-sector collaboration in promoting environmentally responsible AI deployment. Finally, the paper presents practical implications and outlines future research directions to guide the sustainable evolution of AI in healthcare.
1 Aug 2026
Read appraisal →JMIR formative research
Teaching Health Informatics in a Skills-First Era: A Curriculum Mapping Approach Aligned With Workforce Demand
Academic programs in health informatics have been urged to align curricula with evolving workforce needs to better prepare graduates for practice. Such alignment supports workforce readiness, addresses talent shortages, and reduces onboarding burdens for employers. Although employers increasingly seek graduates who can demonstrate applied skills and competencies beyond textbook knowledge, academic programs continue to face challenges in translating workforce demand into curricular design. This Viewpoint is directed at health informatics program faculty and administrators and presents the author's perspective on a skills-first curriculum mapping approach aligned with current industry needs. The author argues that connecting program-level outcomes, course objectives, and learning activities to industry-recognized competencies strengthens workforce readiness while maintaining academic rigor. Drawing on the literature on competency-based education, workforce development, and health informatics training, this paper argues for intentional curriculum design that integrates transferable skills, stackable credentials, and stakeholder engagement. The key messages are as follows: (1) workforce demand should anchor curriculum design in health informatics programs, (2) curriculum mapping is a practical mechanism for operationalizing skills-first alignment, and (3) credential integration and continuous congruency evaluation support long-term program sustainability. This approach positions health informatics education to remain responsive, scalable, and aligned with the rapidly evolving professional landscape.
31 July 2026
Read appraisal →Scientific reports
Adaptive level modification via player skill classification and large language models
Maintaining player engagement in video games requires a careful balance between challenge and player competence. Static difficulty settings fail to account for individual skill variation, while existing dynamic difficulty adjustment systems are limited to tuning low-level game parameters rather than restructuring level content. This paper presents an adaptive level modification framework that personalizes gameplay by continuously inferring player skill and applying targeted structural modifications to level content in real-time. A hybrid behavioral dataset is constructed by combining agent-generated trajectories, produced by Proximal Policy Optimization (PPO) agents, a reinforcement learning approach, trained at three distinct skill levels, with manually collected human gameplay data labeled through clustering. A classifier trained on this dataset categorizes players into expert, normal, and beginner skill levels, achieving an overall accuracy of 97.82%. The classifier output drives a two-stage large language model (LLM) pipeline guided by prompt engineering, which expands a skill-conditioned prompt into a structured modification instruction applied to the current level chunk. A physics-constrained verifier based on a graph-based shortest path method ensures all modified levels remain traversable. Evaluated on Super Mario Bros. levels, the framework achieves a post-modification playability rate of 74.1% at the full-level granularity and 83.5% at the isolated-chunk granularity, closely matching the 80.0% baseline of the original levels.
30 July 2026
Read appraisal →Human vaccines & immunotherapeutics
Artificial intelligence in immune checkpoint inhibitor research: A bibliometric analysis of the landscape
This bibliometric analysis examines the transformative role of artificial intelligence (AI) in immune checkpoint inhibitor (ICI) research. Using VOSviewer, CiteSpace, and Bibliometrix, we analyzed 1,938 publications from the Web of Science Core Collection (2015-2026), revealing a dramatic rise in AI-related ICI studies, led by China and the USA. Key findings demonstrate AI's integration in predicting treatment response, optimizing dosing strategies, and managing immune-related adverse events. Through keyword co-occurrence and citation analyses, we identify critical AI applications including digital pathology and tumor microenvironment characterization. This comprehensive overview provides valuable insights into research trends and emerging frontiers for AI-driven innovations in cancer immunotherapy.
30 July 2026
Read appraisal →PloS one
Evaluating the cultural alignment of multilingual LLMs in typical Japanese workplace scenarios
While current evaluations of LLM cultural alignment predominantly rely on static benchmarks in Western contexts, their ability to navigate generative, high-context socio-pragmatic demands in non-Western environments remains critically underexplored. This study investigates how multilingual LLMs adapt to the Japanese workplace-a stringent stress-test environment characterized by strong high-context communication norms and rigid honorific conventions-using Hofstede's six cultural dimensions as a heuristic framework. We evaluated five state-of-the-art LLMs (LLM-jp, Phi, Llama, Qwen, and GLM) through a large-scale crowdsourced human evaluation. Based on 1,718 valid evaluator sessions, native Japanese raters assessed model outputs to generate a holistic Japanese Workplace Cultural Alignment Score (JWCAS). To dissect the underlying communicative strategies, we paired this with a three-layer diagnostic sub-score analysis (Linguistic Form, Socio-Cultural Values, and Social Action). Our results reveal that leading multilingual models (Phi and GLM) achieved overall JWCAS scores comparable to, or significantly higher than, the native Japanese model (LLM-jp). Crucially, our sub-score analysis demonstrates that holistic evaluation metrics can obscure deep pragmatic deficits: while LLM-jp overfits to surface-level linguistic politeness (Layer 1), it shows critical weaknesses in socio-cultural values (Layer 2) and context-aware social strategies (Layer 3). In contrast, leading multilingual models demonstrate balanced competence across all layers. These findings suggest that true cultural competence requires moving beyond native linguistic mastery, highlighting the necessity of multi-dimensional diagnostic frameworks for cross-cultural AI alignment.
29 July 2026
Read appraisal →Proceedings of the National Academy of Sciences of the United States of America
Race-conscious admissions algorithms and the law.
In recent years, colleges and universities have begun to use machine learning (ML) systems to inform admissions decisions. Meanwhile, in the 2023 case Students for Fair Admissions, Inc. v. President and Fellows of Harvard College, the Supreme Court held that colleges and universities may not make admissions decisions "on the basis of race." These parallel developments-the rise of ML in admissions and the fall of race-based affirmative action-will force educational institutions, and ultimately courts, to confront the difficult question of what it means for ML systems to differentiate "on the basis of race." We begin by mapping the Students for Fair Admissions decision onto different uses of race in predictive AI. We distinguish between "first-order" and "second-order" race consciousness at both the training and predictive phases of machine learning, and we argue that each category of race consciousness raises distinct legal and normative issues. We go on to show that the Students for Fair Admissions decision potentially permits-and even endorses-certain forms of race consciousness. Our analysis is grounded in the observation that the process of developing ML-based systems enables policymakers to calibrate decision making algorithms much more precisely and explicitly in response to specific criticisms of race-conscious affirmative action.
28 July 2026
Read appraisal →PloS one
PAP_NER: A large-scale vietnamese administrative named entity recognition corpus and hybrid deep learning architecture
Named Entity Recognition (NER) is fundamental for automating administrative document processing in digital government systems. However, Vietnamese NLP research faces a critical infrastructure gap: existing datasets focus on generic information extraction (news, medical) rather than domain-specific administrative text. We present PAP_NER, the first large-scale, gold-standard Vietnamese administrative NER corpus comprising 162,801 sentences with 205,807 entity annotations across five entity types critical for e-Government workflows: Agency (CQ), Legal Document (VBPL), Object (ĐT), Datetime (NG), and Quantity (SL). The dataset was constructed through a rigorous human-in-the-loop annotation pipeline, achieving an inter-annotator agreement of κ = 0.85. We demonstrate PAP_NER's value through comprehensive benchmarking of an established hybrid deep learning architecture, PhoBERT-CRF, which couples monolingual Transformer embeddings (PhoBERT) with Conditional Random Fields for structured prediction. PhoBERT-CRF achieves 97.95% Micro F1-score on the PAP_NER test set, significantly outperforming established baselines: BiLSTM+CRF (+2.01%), multilingual XLM-RoBERTa (+2.52%), and pure Transformer approaches (+0.44%). Ablation analysis reveals that the CRF layer provides statistically significant improvements for structurally complex entities (VBPL: + 0.96%, p < 0.05, McNemar's test). We release PAP_NER publicly (DOI: 10.5281/zenodo.18044019) under Creative Commons BY 4.0 license to support reproducibility and enable further research in Vietnamese administrative NLP. This work establishes a foundational dataset and methodology for addressing the Vietnamese government NER gap, with implications for low-resource language NLP research.
28 July 2026
Read appraisal →Medicine
Performance of DeepSeek V3 and ChatGPT-4o in answering esophageal cancer-related questions
Esophageal cancer remains a significant global health issue. ChatGPT-4o and DeepSeek V3 can provide the public with health-related knowledge about esophageal cancer. This study aimed to evaluate the accuracy of DeepSeek V3 and ChatGPT-4o in responding to health knowledge questions related to esophageal cancer. Fifty-two questions related to esophageal cancer were classified into themes of basic knowledge, diagnosis and molecular biology, management of local and locoregional diseases, management of advanced and metastatic diseases, clinical case analysis and patient frequently asked questions (FAQs). These questions were entered into DeepSeek V3 and ChatGPT-4o to obtain responses, and 2 experienced gastroenterologists independently evaluated the accuracy and temporal stability of each response. Overall, the scores of DeepSeek V3 and ChatGPT-4o on all questions were 4 (3-4), and there was no statistically significant difference between the 2 groups. The final scores of DeepSeek V3 in basic knowledge, diagnosis and molecular biology, management of local and locoregional diseases, management of advanced and metastatic diseases, clinical case analysis, and FAQs were 4 (3-4), 4 (3-4), 4 (3-4), 3 (3-4), 4 (4-4), and 4 (4-4), respectively, while the scores of ChatGPT-4o were 4 (3-4), 3 (2-4), 4 (3-4), 3 (3-4), 4 (4-4), and 4 (4-4), respectively. For temporal stability across 2 independent test runs, DeepSeek V3 presented inconsistent responses on 2 questions, and ChatGPT-4o on 1 question; no statistically significant differences were found in overall and subgroup scores between the 2 runs for both models (all P > .05). ChatGPT-4o and DeepSeek V3 showed favorable accuracy and comprehensive responses to most of the 52 esophageal cancer-related questions in this study, but our findings do not confirm their general reliability for esophageal cancer health information in routine clinical or public use.
27 July 2026
Read appraisal →PloS one
Multi-view graph-regularized deep metric subspace clustering network
Multi-view subspace clustering has progressed significantly by using deep neural networks to handle nonlinear data representations. A recent advancement, the Multi-view Self-Expressive Subspace Clustering (MSESC) network, achieves markedly higher computational efficiency by substituting the traditional self-expression layer with a deep metric learning approach. Nevertheless, MSESC still suffers from two notable limitations: it fails to adequately capture the high-order geometric structures inherent in multi-view data, and it lacks effective guidance from the underlying clustering distribution. To overcome these shortcomings, we propose a novel framework termed Multi-View Graph Regularized Deep Metric Subspace Clustering (MVGR-DMSC). The proposed method introduces two key components into MSESC to enhance the discriminability of representations. First, a dual-order graph regularization module is devised to maintain both first-order and second-order manifold structures, thereby allowing the model to capture more complex local geometric relationships. Second, an adaptive view-weighted deep clustering module is incorporated, which employs the Kullback-Leibler divergence to guide representation learning while dynamically adjusting the contributions of different views. Through evaluations on five benchmark datasets, we show that MVGR-DMSC consistently yields better results than several state-of-the-art approaches, including the direct baseline MSESC, in both accuracy and robustness.
26 July 2026
Read appraisal →JMIR human factors
Japanese Health Information Technology Usability Evaluation Scale for Sexually Transmitted Infection-Related Chatbots: Development and Psychometric Validation Study
BACKGROUND: The rapid expansion of mobile technology has accelerated the integration of health applications and conversational AI into clinical and public health practices. To ensure these tools are effective and sustainable, usability evaluations and early user engagement during development are essential. The Health Information Technology Usability Evaluation Scale (Health-ITUES) is a validated and flexible usability assessment instrument that is available in multiple languages and applicable across diverse contexts. However, a Japanese version of this scale has not yet been developed. OBJECTIVE: This study aimed to translate and validate a Japanese version of the Health-ITUES, customized for a sexually transmitted infection (STI)-related chatbot, and to support the usability assessment of emerging mobile health tools in Japan. METHODS: We developed a Japanese version of the Health-ITUES using a chatbot under development as a consultation tool for young women regarding STIs. First, the original scale was customized to reflect the chatbot's specific purpose and intended usage context. Following established translation guidelines, we conducted forward translation from English to Japanese, back translation, expert review, and reconciliation. We then evaluated the reliability and validity of the Japanese version in a sample of 301 young women. RESULTS: The Japanese version of the Health-ITUES demonstrated high internal consistency (Cronbach α=0.85-0.98). Confirmatory factor analysis supported acceptable construct validity (root mean square error of approximation is 0.10, comparative fit index>0.90). Additionally, the Health-ITUES scores showed strong correlations with satisfaction and usage intention for the tool (r=0.779 and 0.797, respectively). CONCLUSIONS: The Japanese version of the Health-ITUES provides initial evidence of reliability and validity in an STI-related scenario among young women and may facilitate more rigorous usability evaluations of mHealth and conversational AI tools in Japan.
24 July 2026
Read appraisal →Behavior research methods
Measuring complex constructs in large-scale text with computational social mixed methods
A growing convergence between social science and machine learning enables, in principle, large-scale analyses of complex social phenomena through text. Yet, approaches leveraging supervised text classification based on human-annotated data for statistical analysis often treat conceptual validity and technical performance as separate challenges, impairing measurement quality. We provide guidelines to bridge this gap in what we call computational social mixed methods pipelines across three stages: data annotation, model training, and statistical analysis. Building on best practices and our own methodological innovations, such as "Iterative Annotation" and "Training on Confident Examples", we address recurring pitfalls like unbalanced training data or stagnant model performance. We also discuss when large language models constitute a viable alternative to transformer-based classifiers. Using a case study on countering online hate, we illustrate how consequently integrating social science and machine learning expertise improves the validity and comparability of computational social science.
24 July 2026
Read appraisal →PloS one
"We'll have to see how it works": An interview study to understand collaborative practices in interdisciplinary artificial intelligence and healthcare research.
Developing artificial intelligence (AI) algorithms for healthcare is a collaborative effort, bringing data scientists, clinicians, patients and other stakeholders together. By understanding AI as 'sociotechnical' where the social and the technical nature of the work and the models are inseparable, we explore the AI development workflow and how stakeholders navigate the challenges and tensions of sharing and generating knowledge across disciplines. We conducted an inductive thematic analysis of 13 semi-structured interviews with participants in early stages of AI-in-healthcare research consortia in the UK. Our findings identify that participants needed to adapt both the tools used for sharing and the information communicated according to their audience, particularly when working with those with a clinical or patient perspective. We identify the novelty of participating in AI research, how AI knowledge is shared, and the inclusion of clinician and patient stakeholder perspectives as key areas within collaborative AI practices in healthcare. These findings highlight that bringing AI into the mix can introduce new obstacles to interdisciplinary work.
24 July 2026
Read appraisal →MedEdPORTAL : the journal of teaching and learning resources
Artificial Intelligence and Large Language Models: A Case-Based, Peer-Teaching Workshop for Preclinical Medical Students
INTRODUCTION: Artificial intelligence tools have rapidly become integrated in health care settings and are quickly affecting medical student education. However, there remains limited formal teaching on such tools in student curricula. This project sought to introduce preclerkship medical students to the basics of large language models and allow them to practice ways to best use these tools to supplement their learning. METHODS: The authors designed, implemented, and evaluated a 60-minute lecture and 100-minute workshop for second-year medical students. The workshop included interactive cases covering various aspects of artificial intelligence use, and some groups were led entirely by student leaders, allowing for peer teaching. RESULTS: One hundred sixty-eight students from Harvard Medical School and Harvard School of Dental Medicine were enrolled in this session. Anonymous pre- and postsession surveys (N = 124 and N = 62, respectively) were collected and compared via unpaired t test assuming unequal variance and showed statistically significant increase in the mean ratings of six 5-point Likert scale questions assessing artificial intelligence-related knowledge/self-efficacy (P < .01) and mixed changes to 5 questions relating to attitudes/behavioral intent. DISCUSSION: Our artificial intelligence and large language model session provides a framework for teaching medical students, early in their educational journey, the basics of these tools. Our session provides interactive exercises to illustrate how best to leverage such tools while also discussing their potential risks. Such education will be important to incorporate into medical student curricula as artificial intelligence technologies grow increasingly common.
23 July 2026
Read appraisal →JMIR mental health
Optimizing Treatment Strategies in the Bipolar Disorder Spectrum With Classical AI Approaches: Systematic Review of Performance, Bias, and Clinical Applicability
BACKGROUND: Bipolar disorder (BD) is a complex and heterogeneous psychiatric condition, characterized by fluctuating clinical courses that affect approximately 1%-2% of the global population in their lifetime. Despite pharmacological advances, treatment response varies significantly among patients, making the identification of individualized treatment strategies a major challenge. Artificial Intelligence (AI), through its classical approaches, has emerged as a powerful tool in precision psychiatry to identify subtle patterns in complex data and inform personalized clinical decisions. OBJECTIVE: The present systematic review aimed to examine the current evidence on classical AI-supported treatment optimization in the BD spectrum. METHODS: The review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines. Four databases (PubMed, Web of Science, Scopus, and Embase) were searched for original studies published after 2015 on the application of classical AI in the treatment of BD in adult patients. Publication bias was evaluated by visual inspection of a funnel plot. The methodological quality, risk of bias, and clinical applicability of the predictive models were assessed using the Prediction Model Risk Of Bias Assessment Tool for prediction models using regression or AI methods (PROBAST+AI; PROBAST+AI Working Group) tool. RESULTS: A total of 35 studies were included and classified into 5 outcome-based categories, including acute symptomatic response, long-term maintenance response, relapse and readmission risk, safety and dose optimization, and brain aging and phenotyping. Acute symptomatic response models performed modestly (pooled area under the curve [AUC] 0.68), while imaging improved accuracy (74%-77%). Long-term maintenance response models showed moderate-to-high performance (pooled AUC 0.80), with biomarker- and cellular-based models reaching 96%-99% accuracy. Relapse and readmission prediction achieved a pooled AUC of 0.71, with digital phenotyping and rule-based methods performing best (AUC 0.85-0.88). Safety and dose optimization models achieved 85%-97% accuracy. Brain aging and phenotyping studies highlighted accelerated brain aging in BD, partially mitigated by lithium, and revealed novel data-driven subgroups. However, 3 studies were considered at high risk of bias due to small sample sizes associated with disproportionately high-performance estimates. An additional study was identified as potentially biased because it lay markedly distant from the funnel plot's confidence line. Finally, the PROBAST+AI assessment revealed a high risk of bias in most studies, primarily due to data analysis limitations, small sample sizes, and lack of external validation. CONCLUSIONS: The adoption of classical AI tools in BD serves as a driver for therapeutic optimization, although current AI tools in BD should still be considered exploratory rather than ready for clinical use. Effective implementation in real-world clinical scenarios requires more robust, transparent, and externally validated models to ensure reliability and generalizability.
22 July 2026
Read appraisal →Journal of medical Internet research
AI-Powered Simulation for Nursing Education: Mixed Methods Systematic Review
BACKGROUND: Traditional simulation-based nursing education is often constrained by high costs, resource intensity, and limited scalability. AI-powered simulations offer dynamic, scalable, and personalized alternatives. However, the empirical evidence regarding their pedagogical effectiveness and learner acceptance remains fragmented. OBJECTIVE: This study aimed to systematically evaluate and synthesize evidence on the effectiveness and learner perceptions of AI-powered simulations in nursing education. METHODS: Following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, we systematically searched 11 electronic databases (PubMed, CINAHL, Embase, Web of Science, Cochrane Library, Scopus, SinoMed, CNKI, Wanfang, VIP, and Google Scholar) for studies published between January 2014 and September 2025. Two independent reviewers performed study selection, data extraction, and quality appraisal using design-specific tools (risk of bias 2 tool [RoB 2; Cochrane Bias Methods Group] for randomized controlled trials [RCTs], Risk Of Bias in Nonrandomized Studies of Interventions [ROBINS-I; Cochrane Bias Methods Group] for nonrandomized studies, Mixed Methods Appraisal Tool [MMAT] for mixed methods, Joanna Briggs Institute [JBI] for qualitative, and Agency for Healthcare Research and Quality [AHRQ] for cross-sectional studies). Quantitative data were synthesized narratively, and qualitative findings were integrated using JBI meta-aggregation. A convergent segregated approach with joint display was used to generate meta-inferences. RESULTS: Nineteen studies involving 1253 participants (primarily prelicensure nursing students, with some interdisciplinary cohorts) were included. AI modalities comprised generative AI/large language models (n=7), AI-driven virtual patients/mannequins (n=5), AI-enhanced virtual/mixed reality (n=5), and chatbots (n=2). Three studies were RCTs, 4 were quasiexperimental with control groups, 3 were uncontrolled pre-post studies, 4 were mixed methods, 4 were qualitative, and one was a cross-sectional survey. Quantitative synthesis showed that evidence from RCTs and controlled quasiexperimental studies indicates significant improvements in cognitive knowledge and affective outcomes, including self-efficacy and communication confidence; however, effects on complex psychomotor skills were inconsistent, with one RCT finding AI-assisted simulation inferior to standardized patient simulation. Findings from uncontrolled designs are preliminary. Qualitative meta-aggregation revealed that learners valued safe, repeatable, nonjudgmental practice environments that reduced anxiety and bridged the theory-practice gap. Persistent challenges included technical frustrations, "robotic" interactions, lack of nonverbal cues, and system instability, collectively constituting an "authenticity gap." CONCLUSIONS: AI-powered simulations show promise for developing foundational clinical reasoning and communication skills in nursing education, though the evidence base is limited by the predominance of uncontrolled designs, reliance on self-reported measures, and absence of longitudinal data on skill retention or clinical transfer. Due to current technological limitations in replicating physical and emotional authenticity, AI should be implemented as a complementary tool alongside traditional simulation methods and clinical placements, rather than as a replacement. Future research should prioritize longitudinal outcomes, standardized competency measures, RCTs with active comparators, and implementation strategies addressing technical barriers.
22 July 2026
Read appraisal →Journal of medical Internet research
Digital Transformation in Health Care: Are We on the Right Track?
Health care digital transformation is gaining increasing attention, despite the observed challenges in its implementation. The envisioned benefits, together with the growing need for better health care, are motivating academia, organizations, regulatory agencies, and governments to develop more effective digital health care solutions. Through extensive debates among the authors, this paper discusses how digital transformation is being conducted in the health care sector. Our discussion relies on concepts from sociotechnical systems theory, categorizing it into 3 social (people, culture, and goals) and 3 technical (processes/procedures, infrastructure, and technology) dimensions. Drawing on both the literature discussed in this paper and our combined academic and practical experience-each author contributing over a decade of work in health care digital transformation-we examined current digitalization efforts from a sociotechnical standpoint. Overall, we argue that both social and technical dimensions present elements that have either encouraged or discouraged the progress of health care digital transformation. The identification of current trends on such (on- and off-track) elements allowed the formulation of 12 propositions across each sociotechnical dimension for future testing and validation. This approach can help establish better government policies, foster private initiatives, and shift regulatory guidelines to support a successful digital transformation in health systems. Finally, from a research perspective, we outline some opportunities for further interdisciplinary investigation in the field, promoting advances in the understanding of health care digital transformation.
22 July 2026
Read appraisal →Scientific reports
The role of AI in combating misinformation: leveraging text mining and social networking analysis
Misinformation on social media can be a severe threat to social trust, safety, and health of the population, especially in times of an epidemic like the Monkeypox outbreak. This study specifically focuses on a hybrid RoBERTa-GRU architecture designed to capture both contextual semantics and temporal dependencies in social media discourse. This research presents how the combination of text mining and social network analysis enables Artificial Intelligence (AI) to support misinformation detection. The proposal of a hybrid architecture that integrates RoBERTa and GRU-based embeddings in a contextual fashion and GRU-based modelling of sequential patterns helps identify and substantiate misinformation in social media posts. Based on a curated X (formerly Twitter) dataset consisting of Monkeypox posts (5787 posts), the model provided state-of-the-art results, with ROC-AUC 0.9979 and Cohen's kappa 0.9887; standalone baselines were also surpassed. Results reveal that the proposed transformer-RNN hybrid effectively captures both semantic depth and temporal relationships in misinformation detection tasks. In addition to performance, the paper addresses the limitations of dataset bias, multilingual constraint issues, and scalability as related to cross-linguistic applicability, multimodal study, and performance in real-time and resource-limited systems. The study adds value in this emerging body of knowledge on AI-driven social media analytics by offering practical guidance for mitigating health-related misinformation online.
22 July 2026
Read appraisal →JMIR formative research
Development of a User-Informed Decision Aid for Contraceptive Decision-Making Among Adolescents and Young Adults: Qualitative Study
BACKGROUND: Adolescents and young adults seeking contraceptive care face many considerations related to differences in contraceptive indications and knowledge, which are often overlooked and can lead to contraceptive nonadherence or nonuse as well as adverse health outcomes. OBJECTIVE: This study aimed to develop a digital decision aid that meets the contraceptive decisional needs of adolescents and young adults from diverse backgrounds. METHODS: We developed a web-based decision aid using a user-centered design framework and the International Patient Decision Aid Standards. The design and development process was informed by a literature review and consultations with scientific experts, health informatics specialists, clinicians experienced in adolescent reproductive health, and a diverse group of adolescent and young adult advisors. We gathered feedback on the decision aid's content and functionality from clinicians, adolescents, and young adult stakeholders during focus group interviews conducted across 2 user testing cycles. Each focus group was recorded and transcribed, and the data were analyzed using a focus group guide to identify key attributes, patterns, and perspectives among users. Two qualitative researchers used rapid qualitative analysis to explore and summarize findings across 4 key domains, which contributed to the refinement of the decision aid content and the improvement of its functionality. RESULTS: Twenty-four clinicians, adolescents, and young adult participants from diverse backgrounds shared their perspectives on the decision aid's content, relevance, design, and usability across 2 user testing cycles conducted from February 2023 to June 2024. The decision aid includes a survey, a decision algorithm that generates a summary of contraceptive method recommendations, infographics, and a health care provider summary view. The decision algorithm applies a weighted scoring system ranging from +1 or -1 to +10 or -10 for each method, reflecting the user's primary indication for seeking contraception, contraceptive use preferences, and relevant health history. CONCLUSIONS: This approach allowed us to capture rich perspectives from a diverse group of stakeholders that accounted for the unique contraceptive decision-making needs of adolescents and young adults, resulting in a functional, youth-informed decision aid prototype, MyPlanMyChoice, ready for pilot and feasibility evaluation in a clinical setting.
22 July 2026
Read appraisal →JMIR cancer
Multiturn Large Language Model-Based Conversational Agents for Patients With Cancer and Caregivers: Scoping Review
BACKGROUND: Large language model (LLM)-based conversational agents are increasingly used in health care, yet their capacity to support genuine multiturn dialogue remains underexplored. In oncology, where patients and caregivers experience complex informational and emotional needs throughout the disease trajectory, conversational agents may support information provision, symptom consultation, and emotional assistance. However, research specifically examining multiturn conversational agents designed for patients with cancer and informal caregivers remains limited. OBJECTIVE: This scoping review aimed to map the research landscape of LLM-based multiturn conversational chatbots developed for patients with cancer and informal caregivers, focusing on system design, intervention purposes, evaluation approaches, safety considerations, and transparency of LLM-related components. METHODS: This scoping review followed the Joanna Briggs Institute methodology and PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines. Six databases-PubMed, Embase, Scopus, Web of Science, CINAHL, and PsycINFO-were searched for studies published between January 2022 and January 2026, with supplementary searches conducted in IEEE Xplore Digital Library and ACM Digital Library in May 2026. Studies were included if they described LLM-based chatbots designed for patients with cancer or informal caregivers that supported multiturn conversational interaction. Two reviewers independently conducted the study selection and data extraction. RESULTS: Eight studies met the inclusion criteria. Most studies focused on prototype development, with limited research evaluating clinical outcomes. ChatGPT-based models were the most commonly used LLMs, and retrieval-augmented generation techniques were applied in several studies. Chatbots were primarily designed for emotional support or information provision. Evaluation approaches varied widely, including response quality, psychological outcomes, and user experience. However, no studies evaluated interaction-level characteristics such as conversational continuity or context retention, and only 2 studies reported any conversational memory mechanism. Reporting on safety risks, mitigation strategies, prompt design, model parameters, and adherence to LLM reporting guidelines was often limited or absent. CONCLUSIONS: This scoping review identified only 8 studies on LLM-based multiturn conversational chatbots for patients with cancer and informal caregivers. The field remains at an early stage, characterized by prototype-oriented development, heterogeneous design and evaluation approaches, and inconsistent safety and transparency reporting. Future development should prioritize genuine conversational capability, safety management, and transparent reporting.
22 July 2026
Read appraisal →Injury prevention : journal of the International Society for Child and Adolescent Injury Prevention
Natural language processing tool for extracting information about opioid overdoses in the USA from case narratives in the violent death reporting system.
BACKGROUND: Improving the infrastructure for drug overdose surveillance is critical for identifying new threats and responding to emerging trends. We aimed to develop a prototype tool using the principles of natural language processing that can extract information from the death records of drug overdose victims. METHODS: Data were obtained from the Violent Death Reporting System on drug overdose deaths. Narratives were manually labelled for 12 attributes of interest, totalling 82 labels about the circumstances of the overdose. Narratives were passed through the 'Excel Extractor' to identify and extract a target phrase and subsequently map the extracted phrase to predetermined code values. The output from the Excel Extractor was compared with manually labelled data to determine accuracy. Performance was compared against multiple machine learning models. RESULTS: The Excel Extractor performed well across the attributes of interest, achieving an F1 Score over 0.8 on nine of the 12 attributes. The Excel Extractor was the highest performing model on seven of the 12 attributes. The Excel Extractor achieved an F1 Score of 0.8 or higher on 46 of 82 (56%) of the labels, and a score of 0.9 or higher on nearly one-third (25 out of 82) of the labels. CONCLUSION: This work demonstrates it is feasible to develop a spreadsheet-formula-based natural language processing tool to accurately extract information about drug overdose deaths from narratives; for most attributes, a rule-based search performs well or better than deep learning. The Excel Extractor has the potential to streamline data abstraction for epidemiologists gathering data about drug overdose deaths.
21 July 2026
Read appraisal →European journal of clinical pharmacology
Clinical decision support systems for polypharmacy optimization in older patients: a narrative review
PURPOSE: Multimorbidity and polypharmacy are increasingly prevalent in the older population and are associated with a higher risk of potentially inappropriate medications (PIM), drug-drug interactions (DDI), and adverse drug reactions (ADR). Although medication review (MR) and deprescribing are effective strategies, their manual implementation can be complex, time-consuming, and prone to clinical variability. Clinical Decision Support Systems (CDSS) offer advanced digital solutions to optimize polypharmacy by analyzing multidimensional clinical data and generating personalized recommendations. METHODS: A narrative review was conducted to identify and compare the main CDSS developed for the polypharmacy management, MR and deprescribing in older adults and patients with multimorbidity. Systems were classified as manual, hybrid, or automated according to their data acquisition modalities. Operational characteristics, integration into clinical workflows, decision-support functions, generated outputs, and available validation evidence across different healthcare settings were assessed. Owing to the narrative nature of the review and the heterogeneity of the included evidence, no formal risk-of-bias assessment or certainty-of-evidence evaluation was performed. RESULTS: Manual CDSS require direct data entry by clinicians and are associated with a high operational burden. Hybrid systems combine automatic data acquisition with manual integration, balancing efficiency and clinical oversight. Automated systems, integrated into electronic health records (EHR), provide real-time decision support with minimal human intervention. Considerable heterogeneity was observed across identified platforms in terms of automation, implementation characteristics, and stage of validation, with evidence ranging from development and feasibility studies to observational analyses and randomized controlled trials (RCT). CONCLUSION: CDSS represent promising tools for safer and more effective management of polypharmacy in complex patients. Advanced integration into clinical workflows and systematic use of multidimensional data may enhance their impact. However, the heterogeneity of available systems and the variability in their level of clinical validation highlight the need for comparative studies, pragmatic trials, and real-world implementation evaluations. Such studies are necessary to clarify the impact of different CDSS models on prescribing appropriateness, medication-related risks, patient outcomes, and long-term sustainability within routine healthcare settings.
20 July 2026
Read appraisal →JMIR cardio
Mobile Apps for Heart Rate Variability: App Store Search and Content Analysis
BACKGROUND: Heart rate variability (HRV) is a noninvasive indicator of autonomic nervous system activity that is increasingly used for health and performance monitoring. Digital and mobile technologies are increasingly providing opportunities for remote HRV monitoring outside of laboratory-based settings. OBJECTIVE: This study aimed to describe the landscape of mobile apps that measure, analyze, and provide feedback on HRV, with a focus on how HRV is measured, analyzed, interpreted, and communicated to users. A secondary aim was to assess the transparency of these apps, including the extent to which they disclose the evidence underpinning their HRV metrics and feedback. METHODS: This study was an app store search and content analysis. Searches were conducted in the Google Play Store and Apple iTunes Store. Apps were eligible for inclusion if they had functionality to record, analyze, or provide feedback on HRV and were available in English. Data were extracted from app descriptions, screenshots, websites, and, where necessary, contact with developers. Data were extracted on app metadata (developer, release and update dates, and pricing), alongside information about HRV measurement, analysis, and feedback. This included the type of sensor used; HRV measurement characteristics (sensor placement, recording duration, and body position or standardization procedures); methods to calculate and interpret HRV (ie, metrics derived and how they were interpreted for users); and additional app functionality such as reminders, the ability to log self-reported stressors, and the type of feedback or guidance provided based on HRV. We used previously published criteria for assessing the quality of information on the internet, which included authorship, scientific attribution, currency of updates, and data privacy. RESULTS: Of 746 apps identified, 206 met eligibility criteria. Of these, 132 were primary measurement apps, 59 were aggregators, and 15 were hybrid. Photoplethysmogram was the most common sensing modality (n=117, 56.8%), followed by multiple sensors (n=60, 29.1%). Full data extraction across app metadata and HRV measurement and analysis data was only achievable for 93 (45.1%) apps, representing a transparent subset with sufficient available information for content analysis. The most commonly reported HRV metrics were root mean square of successive differences (n=51) and SD of normal-to-normal intervals (n=48), while frequency-domain power (n=22) and low frequency to high frequency ratios (n=15) were less common. Most apps presented data as personalized trends or individualized ranges (76/93, 81.7%), emphasizing user-specific context rather than isolated values. Although 86% (80/93) offered contextual guidance (eg, readiness or recovery scores), many relied on proprietary algorithms that were not transparently described, limiting independent assessment of how these scores were derived and validated. CONCLUSIONS: Consumer HRV apps are widely available but vary considerably in how data are collected, processed, and contextualized. While many offer personalized trends and guidance, methodological transparency is often limited, particularly regarding the proprietary algorithms underlying the feedback scores.
20 July 2026
Read appraisal →Journal of medical systems
Recent Advances in AI for Automated ICD Coding: A Systematic Literature Review
International Classification of Diseases (ICD) codes enable correct billing, insurance reimbursement, and healthcare analytics. However, manual coding is time-consuming, expensive, and error-prone, creating bottlenecks in clinical workflow and limiting scalability. Artificial intelligence (AI) has emerged as a promising solution for automated ICD code assignment from unstructured clinical text. This systematic review explores the current state of automated ICD coding research, examining models applied to diverse clinical documents including discharge summaries, electronic health records, nursing notes, and pathology reports. Following PRISMA guidelines, we searched six databases for studies published between 2019 and 2024, selecting 54 relevant studies from 4,280 initial citations. Our analysis reveals the use of diverse datasets, preprocessing techniques, and feature extraction methods, alongside a clear evolution from traditional machine learning to deep learning approaches, with substantial architectural diversity across convolutional, recurrent, transformer, and hybrid models. Performance varies considerably across dataset configurations, with models achieving higher accuracy on frequent code subsets compared to full label spaces. However, critical gaps persist: overreliance on single-language, single-institution datasets limits generalizability; difficulties in predicting rare codes remain unresolved; lack of model interpretability undermines clinical trust; and inconsistent evaluation protocols hinder meaningful comparison. To address these challenges, we propose a 5P evidence-grounded research agenda: Population Diversity, Performance Robustness, Prediction of Rare Codes, Provenance Transparency, and Practical Integration. These findings underscore AI's potential to transform ICD coding while highlighting the need for standardized benchmarks, rigorous external validation, multilingual datasets, and explainable architectures to enable equitable and effective deployment in real-world healthcare systems.
19 July 2026
Read appraisal →Medicine
Exploring the role of generative artificial intelligence in enhancing clinical skills training: A bibliometric analysis
BACKGROUND: Traditional clinical skills training faces challenges such as limited standardized patient resources, high costs of simulation equipment, and restricted opportunities for repeated practice. Generative artificial intelligence (GAI), particularly large language models, has attracted increasing attention as a potential tool for simulation, feedback, and adaptive learning in medical education. However, the research landscape and thematic development of GAI in clinical skills training remain insufficiently mapped. METHODS: A bibliometric analysis was conducted using literature retrieved from the Web of Science Core Collection from January 1, 2011 to April 1, 2025. VOSviewer, the Bibliometrix R package, and CiteSpace were used to analyze publication trends, country and institutional contributions, journal distribution, collaboration networks, keyword co-occurrence, co-cited references, and citation bursts. RESULTS: A total of 322 publications were included. Research activity remained limited before 2023 but increased rapidly thereafter. The United States contributed the largest number of publications, followed by China and India. Major contributing institutions included the National University of Singapore, Gazi University, and Nova Southeastern University. Frequently publishing journals included JMIR Medical Education, Medical Teacher, and BMC Medical Education. Keyword and thematic analyses showed that current research attention is mainly concentrated on ChatGPT, large language models, natural language processing, clinical reasoning simulation, personalized learning, virtual patient interaction, and ethical governance. CONCLUSIONS: Research on GAI in clinical skills training is in an early but rapidly expanding stage, with growing scholarly attention to language-model-driven educational applications and related ethical issues. The bibliometric findings reflect research activity, knowledge structure, and thematic priorities rather than direct evidence of educational effectiveness. Future studies should adopt rigorous empirical designs and standardized evaluation frameworks to assess the effectiveness, safety, and appropriate boundaries of GAI applications in clinical skills training.
19 July 2026
Read appraisal →Philosophical transactions. Series A, Mathematical, physical, and engineering sciences
On the next generation for neuromorphic computing and neuromorphic AI
As we learn more about the inner cognitive workings of the brain's information processing and decision-making behaviours, this naturally leads us to consider alternative and additional ways to process information, from chips and architectures through the generation of insights and decisions. In turn, this suggests some options that might respond to challenges that are not addressable by the present state-of-the-art. By their nature, however, they may develop some common features and idiosyncrasies that are associated with human cognition, such as individual expertise and blind spots, illusions, systematic errors and transient mindsets, as well as evolutionarily advantageous fast-thinking facilities, which are applicable within novel and data-poor circumstances. We discuss some of the lessons learned from the reverse engineering of very large-scale neuron-to-neuron simulations (1B neurons) within cortex-like, network-of-networks, architectures. We identify some elements of the dynamical behaviour of the inner sub-networks (neural columns) that are not exhibited by present-day neuromorphic chips, owing to conceptual and design limitations. We describe a novel mathematical framework that might encompass human cognitive processing alongside various future neuromorphic processing concepts. We also identify certain elements of human cognition, reasoning and performance that present-day chips and present-day artificial intelligence (AI) simply cannot fully emulate (match to a high standard, in some artificial way), or simulate (achieve in the same way). We discuss how these aspects might catalyse some new fields of development for both processors and AI methodologies. In short, we discuss how future research and development will respond in radical ways that are precluded by most present-day technologies. This article is part of the theme issue 'Safe, secure and robust AI for safety-critical systems'.
18 July 2026
Read appraisal →PloS one
A hybrid deep learning model for user story effort estimation
Accurate effort estimation of user stories is a key challenge in agile software development due to both subjectivity and the complexity of natural language requirements. This paper proposes a hybrid Deep Learning (DL) model for data driven effort estimation using large scale textual data and advanced semantic modeling. One of the significant contributions is the development of a dataset of 6,956 user stories which was collected from many heterogeneous sources and then meticulously cleaned and refined to 4,079 high quality instances, by using a systematic preprocessing and expert validation. Following a Design Science Research (DSR) methodology, a hybrid model integrating a pre-trained Bidirectional Encoder Representations from Transformers (BERT) encoder with a Long Short-Term Memory (LSTM) is developed to capture both contextual semantics and sequential dependencies in user story description. The DL model is evaluated against multiple Machine Learning (ML) baselines using a robust multi-metric framework. Experimental findings show the superior performance of the proposed model with a Mean Absolute Error (MAE) = 0.6481, Root Mean Square Error (RMSE) = 1.4559 and = 0.6581, which is a huge improvement over the conventional methods. To ensure practical relevance in discrete Scrum planning, the continuous model outputs were mapped to the standard Fibonacci sequence, achieving a classification accuracy of 72%. To guarantee the consistency of performance improvements, the statistical validation is done with the Wilcoxon Signed-Rank Test to show that the improvements are significant (p < 0.05). Moreover, the model is operationalized as web based decision support system to enable real time estimation in agile software development. The results reveal the efficacy of using large scale curated datasets alongside hybrid DL model to reduce estimation bias and boost predictive quality to deliver a scalable and strong solution for intelligent agile project management. Future work will focus on integrating eXplainable AI (XAI) techniques and validating the model across real world industrial datasets to further enhance transparency and generalizability.
18 July 2026
Read appraisal →Studies in health technology and informatics
Digital Twins for Decision Making: Combining Ambulance and ED Facility Resources
Emergency department (ED) walk-ins and ambulance arrivals are often modelled independently, overlooking their interdependence and the impact of fixed treatment space (TS) capacity. An established discrete event simulation algorithm has been tailored as the basis of a digital twin (DT). Driven by regular streams of incoming ED presentations, patient off-stretcher time (POST), ambulance bay and TS occupation across a 24-hour period contingent on patient flow and ED TS capacity was simulated. The simulation was calibrated using 19,570 ambulance records and data from a 250-bed hospital (2021-2022). Scenarios of five ambulance bays with 43, 38, and 34 TS capacities were investigated, revealing patterns of TS depletion and recovery, including cases where recovery failed within 24 hours, underscoring the sensitivity of patient flow to resource constraints and the value of DT-based simulation for capacity planning.
17 July 2026
Read appraisal →Health care management science
Improving hard-to-place kidney allocation: A machine learning approach to center ranking
Kidney transplantation is the preferred treatment for end-stage renal disease, yet donor scarcity and inefficiencies in allocation systems create major bottlenecks, resulting in prolonged wait times and alarming mortality rates. Despite the severe shortage of donor kidneys, timely and effective interventions to prevent non-utilization of life-saving organs remain limited. Expedited out-of-sequence placement of hard-to-place kidneys to centers with a high likelihood of acceptance has been recommended in the literature as a strategy to improve placement success. However, in practice, this process remains nonstandardized and relies heavily on the subjective judgment of decision-makers. We propose a data-driven, machine learning-based ranking policy for out-of-sequence allocation of hard-to-place kidneys that prioritizes transplant centers using predicted center-level acceptance probabilities. Using national deceased-donor and kidney-offer data, we construct a unique offer-level dataset with donor- and center-specific features. We also employ machine learning interpretability tools to provide insight into the factors influencing kidney allocation decisions. Our analysis demonstrates that the proposed policy can reduce the average number of centers considered before placement by fourfold for all kidneys and tenfold for the subset of hard-to-place kidneys. These results highlight the potential of the proposed framework to improve the efficiency of expedited placement and support more timely utilization of hard-to-place kidneys.
17 July 2026
Read appraisal →International journal of medical informatics
A scoping review on conversational AI in mental health: A human-centered perspective.
BACKGROUND: Conversational AI offers scalable mental health support, with large language models (LLMs) enabling personalized interactions. Human-centered design is critical in this domain, yet a comprehensive synthesis from this perspective is lacking. This review maps conversational AI research in mental health across the patient journey and develops a human-centered taxonomy to guide future design. METHODS: Following PRISMA guidelines, we conducted a comprehensive search across fifteen multidisciplinary databases. We systematically analyzed the literature across six dimensions: research foci, mental disorder types, target populations, AI technologies, data sources, and evaluation metrics. A consensual taxonomy research method was employed to develop a human-centered design framework. RESULTS: Of 10,293 identified records, 677 studies met the inclusion criteria. Analysis reveals a marked increase in publications since 2020, predominantly from computer science (449 studies), followed by medicine (148) and social sciences (80). Research is skewed toward detection (23%) and intervention (66%) stages, with prevention (8%) and maintenance (3%) receiving less attention. Mood, anxiety, and stress-related disorders are the most investigated conditions. LLMs have emerged as the predominant AI technology, particularly within intervention and maintenance stages. Data sources continue to rely heavily on text-based inputs, with multimodal approaches still limited in adoption. Evaluation metrics vary significantly by discipline, reflecting limited cross-disciplinary integration. Through thematic synthesis, we developed a human-centered taxonomy comprising four primary dimensions: Emotional Sensitivity to Users, User-Centric Interaction Design, Human-AI Collaboration and Capability Enhancement, and Ethics and Accountability, with a total of thirteen sub-dimensions. CONCLUSIONS: This review provides a comprehensive, human-centered mapping of conversational AI research in mental health across the patient journey. Critical gaps remain in stage coverage, disorder diversity, population inclusivity, multimodal data integration, and interdisciplinary evaluation. The proposed taxonomy offers a structured framework to align AI development with human-centered principles, fostering empathetic, ethical, effective, and equitable mental health support.
16 July 2026
Read appraisal →PloS one
A dual-stream deep learning architecture for business impact scoring and alert escalation
Modern network monitoring systems generate massive volumes of telemetry data, yet most existing anomaly detection models fail to prioritize alerts according to their operational urgency and business impact. This limitation results in delayed incident responses and inefficient alert management in Network Operations Centers. To address this gap, this study proposes a Dual-Stream Predictive Alert Escalation Framework that integrates temporal failure pattern learning with business impact-aware alert prioritization. The proposed architecture consists of two key components: a bidirectional temporal encoder for modeling multivariate Key Performance Indicator (KPI) time-series data, and an auxiliary severity encoder that captures contextual metadata related to operational risk and service criticality. The outputs of these two learning streams are combined through an attention-based fusion mechanism, and a Business Impact Scoring (BIS) layer generates impact-weighted escalation decisions for proactive incident management. Experimental evaluations using real-world KPI datasets and the AI4I_2020 predictive maintenance dataset demonstrate the superior performance of the proposed framework compared to baseline methods such as LSTM, GRU, CNN-LSTM, and BiLSTM-VAE. On the combined multivariate KPI dataset, the model achieved a precision of 0.94, a recall of 0.92, and an F1-score of 0.93, along with a PR-AUC of 0.95 and a ROC-AUC of 0.94. Under impact-aware evaluation, the framework attained the highest Impact-Weighted F1 (IW-F1) of 0.85 and BIS accuracy of 0.88, resulting in an estimated 31.6% reduction in operational costs through earlier and more accurate escalation of critical events. The suitability of the selected datasets is justified by their complementary roles: publicly available KPI time-series datasets represent real-world network telemetry behavior, while the AI4I_2020 dataset provides structured severity and operational context, enabling joint evaluation of failure prediction accuracy, escalation timeliness, and business impact modeling. By prioritizing alerts based on impact-aware severity rather than raw anomaly scores, the proposed framework directly supports operational cost reduction through earlier mitigation and improved decision-making in network operations. The proposed approach bridges the gap between anomaly detection and intelligent alert management by incorporating business relevance into predictive modeling. This dual-stream architecture offers a scalable and proactive solution for AIOps-driven network reliability and automated service resilience.
14 July 2026
Read appraisal →BMC health services research
Computational forecasting using Swedish data in the vaccination phase of the COVID-19 pandemic: a systematic literature review deliberating modelling relevance for public health and healthcare.
BACKGROUND: The benefits of computational forecasting in the later phase of the COVID-19 pandemic when vaccines and clinical pharmaceutical interventions were available have seldom been assessed. We aimed to evaluate computational forecasting research applied to Swedish populations in the vaccination phase of the pandemic. METHODS: A systematic search was performed on March 8, 2024 in the electronic databases PubMed, Scopus, Cochrane library, Embase, Love platform and Epistemikos. An updated version of the Risk of bias Opinion Tool (ROBOT) was used to assess the quality of evidence reported in the papers identified in the search. The articles fulfilling the quality criteria were assessed for suitability for meta-analysis. Data were extracted from the selected articles for synthesis of characteristics, and a thematic analysis was used for a qualitative synthesis of the contents. RESULTS: Of 2034 unique publications identified in the database search, 6 articles satisfied the selection and quality criteria. Variability in the reporting of forecasting performance results was found to make a quantitative meta-analysis of forecast performance infeasible. The data synthesis showed that statistical modeling using Bayesian calibration was the most common methodological approach. No external model validation was reported, but 5/6 articles included internal model corroboration data. The primary theme resulting from the qualitative synthesis of article content was design or refinement of computational models with demonstration of model use in health service practice as a secondary theme. None of the articles referred to health service policymaking as the primary research context. CONCLUSION: Computational forecasting research using Swedish population data from the vaccination phase of the COVID-19 pandemic was deployed in a model design context. While methodological knowledge was developed, most of the research was not initiated to solve the public health and healthcare problems at hand. Our results indicate that the alignment between computational forecasting research and policymaking needs in the vaccination phase of pandemics can be enhanced.
13 July 2026
Read appraisal →JMIR medical informatics
Natural Language Processing Applied to Psychiatric Clinical Notes: Scoping Review
BACKGROUND: Psychiatric clinical notes in electronic health records (EHRs) provide rich longitudinal information that can support clinical decision-making. Using historical medical data can enable earlier identification of mental illness, better characterization of disease trajectories, and more personalized treatment planning. Natural language processing (NLP) transforms these unstructured notes into analyzable representations for research and care. OBJECTIVE: This study aims to systematically summarize NLP methodologies for psychiatric clinical notes, compare major modeling paradigms and application areas, and highlight emerging large language model (LLM) trends, key challenges, and future research directions. METHODS: Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines, a literature search was conducted for articles on NLP methods based on psychiatric clinical notes published from January 2021 to December 2025 in Ovid MEDLINE, Ovid EMBASE, PubMed, Scopus, Web of Science, the ACM Digital Library, and ScienceDirect. This scoping review analyzed NLP methods applied to psychiatric clinical notes, focusing on major trends, identifying suitable features for traditional machine learning (ML)-based models, applications of pretrained language models (PLMs), and key challenges. Approaches were categorized as rule-based, traditional ML, hybrid, deep learning (DL), and LLM-based methods across information extraction and text classification tasks. RESULTS: In total, 101 studies were eligible for inclusion. Rule-based methods (n=36) and hybrid approaches (n=34) remained the most widely used techniques, largely favored for their interpretability in handling nuanced, subjective clinical notes. These were followed by DL (n=15), traditional ML (n=10), and LLM-based approaches (n=6). Traditional ML studies relied heavily on engineered features, which could be grouped into 5 broad categories: domain knowledge features, lexical and statistical features, vector-based semantic features, emotion-related features, and temporal features. PLMs improved performance mainly through domain adaptation and task-specific fine-tuning, enhancing the handling of psychiatric language, medical terminology, and clinical note structure. LLM-based studies, although still limited in number, indicated a growing shift toward generative and reasoning-based applications. CONCLUSIONS: Hybrid NLP approaches remain dominant, combining domain rules with ML for extraction and classification. DL approaches continue to advance, with domain adaptation supporting medical terminology and clinical semantics. LLMs may further automate complex workflows via zero-shot capabilities and reasoning, alongside growing interest in temporal modeling and multimodal integration. Key future needs include improved generalizability across institutions, privacy protection, and careful attention to ethical implications in clinical deployment.
13 July 2026
Read appraisal →JMIR mental health
Digital Mental Health Research Priorities, Revisited for the AI and Large Language Model Era
Digital mental health has become an established part of mental health care, but the rapid arrival of large language models and other artificial intelligence (AI) tools has refocused attention on the evidence needed to guide the field. This editorial updates the research priorities articulated by JMIR Mental Health in 2023, while reaffirming their emphasis on equity, replicability, privacy, efficacy, and engagement. While the importance of these priorities has not changed in recent years, the urgency with which they must now be applied has. As digital tools become more clinically consequential, research must move beyond demonstrating that a technology is feasible, usable, or novel. The field now needs studies that clarify how these tools work, for whom they are beneficial, under what conditions they may cause harm, and how they can be ethically integrated into care. We call for research that is transparent about the technologies being studied, grounded in meaningful clinical questions, attentive to safety, and designed to produce knowledge that remains useful as specific products and models change.
13 July 2026
Read appraisal →Journal of medical Internet research
Evolution of Regional Information Infrastructures Integrating Health and Social Care in Scotland: Qualitative Study
BACKGROUND: Expectations of integrating health and social care providers have driven the development of digital solutions aimed at overcoming interoperability challenges and ensuring access to information needed for integrated care across fragmented services. However, challenges persist in aligning diverse coding practices, heterogeneous data-sharing mechanisms, and stakeholder needs. OBJECTIVE: We examine how expectations of interoperability and integrated care have shaped the growth of regional information infrastructures in Scotland, using the Key Information Summary (KIS), a summary record that shares key patient information from general practitioner records with out-of-hour services, ambulance services, hospitals, social workers, and caregivers across multiple care settings, as a case study. METHODS: This qualitative study examined the development, implementation, and adoption of KIS in Lothian, Scotland, across health and care settings, where it has been in use for 13 years. Multisited ethnography was used to understand how technology design, implementation, and adoption were shaped by social, organizational, cultural, and political factors. Data were collected through interviews with users, vendors, and implementers; observations of technology use and multidisciplinary team meetings; and documentary analysis of policies, user guides, and internal reports. A hybrid analytical approach was applied: the Technology, People, Organization, and Macroenvironment framework guided initial coding, while the sociology of expectations and information infrastructure theory were used inductively to trace evolving visions of integration, and the long-term development of regional information infrastructure. RESULTS: Data included 54 qualitative interviews, 20 hours of observation, and 59 documents collected between April 2024 and March 2025. Findings illustrate how information infrastructures for integrating health and care providers evolved through successive concerted efforts, conceptualized as waves. Three waves were identified, each characterized by attempts to interlink disparate information systems used by various health and care providers. The first wave focused on linking health care providers by developing networks and architectures required for sharing clinical information, which later supported the development and sharing of KIS. Subsequent waves sought to interlink information systems used by health care providers with those used by local authorities and social care providers. In the absence of shared data standards across these sectors, interoperability was achieved by extending the existing health care-centric infrastructure to different social care settings through workarounds such as providing proxy access to hospital systems and secure emailing networks. CONCLUSIONS: This work illustrates how regional information infrastructures for integrated care evolve through orchestrated waves of change. Some expectations for change required coordinated, system-level action, such as setting up standards, networks, and architecture, while others were realized through local adaptations. Integrating health and care providers through digitalization is a long-term process requiring sustained coordination, with progress often occurring through incremental, local extensions. Policies must support adaptive, long-term coordination, balancing system-level initiatives with local adaptations to achieve meaningful integration.
12 July 2026
Read appraisal →PloS one
NLOS/LOS identification with LightGBM ensemble
Non-Line-of-Sight (NLOS)/Line-of-Sight (LOS) identification is crucial to accurate Ultra-Wideband (UWB) positioning. The current Machine Learning solutions to this problem have either too many parameters to tune or too simple features to input, which lead to unsatisfactory performance. To address this issue, this paper proposed a novel binary classifier called LightGBM Ensemble which integrates multiple LightGBMs in parallel with multi-scale patch extraction. The heterogeneous LightGBM ensemble architecture boosts the prediction power of individuals. The multi-scale patch extraction scheme extracts informative features from time-frequency domains. Extensive experiments on an open-source dataset were conducted to evaluate the proposed approach, which proves its superior classification performance and generalization performance with feasible complexity compared to the state-of-the-art Deep Learning and Decision Trees methods.
12 July 2026
Read appraisal →JMIR mHealth and uHealth
Perceived Sensitivity of Sensor-Based Digital Health Data: Qualitative Interview Study
BACKGROUND: Digital health tools are increasingly used in mental health care to passively collect patient data and analyze health status outside of clinical settings. While technologies such as digital phenotyping, affective computing, and computational behavioral analysis offer new insights into symptom manifestation in daily life, they generate large volumes of potentially sensitive data that raise significant data privacy concerns, requiring high levels of patient awareness and consent. Empirical research is lacking on stakeholder understandings toward the sensitivity of these data and expectations for data stewardship, perspectives that are critical for developing robust informed consent and data protection policies for digital health data use. OBJECTIVE: This study aimed to explore key stakeholder perspectives on the sensitivity of computer perception (CP) data, trust in existing data protections, willingness to share CP data externally, and desire for transparency of CP data transactions outside of the clinical space. METHODS: As part of a larger, multisite study, we conducted qualitative interviews (n=40) via Zoom (Zoom Communications, Inc) with 20 adolescents (aged 12-17 years) familiar with CP tools and their caregivers (n=20). Interviews consisted of a series of open-ended questions regarding stakeholders' perspectives on privacy, data security, and the use and exchange of CP data. We developed a qualitative codebook to identify and label thematic patterns in responses to questions addressing the topics above, using thematic content analysis to identify themes inductively. Each interview was coded by merging work from at least two separate coders, and several team members contributed to qualitative analysis. RESULTS: Most adolescents and caregivers viewed CP data as highly sensitive and expressed a reluctance to share these data beyond their clinical teams. While many participants expressed trust in existing data protections to protect CP data, they often misunderstood or overestimated the extent of protections to safeguard CP data. CONCLUSIONS: Our findings underscore the critical need for clear and effective patient communication and education about the risks, benefits, and protections associated with CP data through informed consent protocols. To promote greater transparency, understanding, and trust, we recommend 5 strategies: educating patients about data protection; studying secondary data exchange and reidentification risks; strengthening transparency regulations; improving data traceability mechanisms, such as distributed ledger technologies, to enhance data traceability and auditability; and adopting dynamic consent models.
11 July 2026
Read appraisal →Nature communications
Empowering clinical trial design with agentic intelligence and real-world data
Clinical trial design (CTD) is a time-consuming process that requires substantial domain expertise. Large-scale real-world data (RWD), such as electronic health records (EHR), encodes practice-based evidence that is of tremendous value to CTD. In recent years, many machine learning methods have been developed to extract such real-world evidence (RWE) from the RWD to inform CTD, but they still need to be communicated with the domain experts extensively in an iterative manner to be further refined and ultimately useful. In this paper, we introduce EmulatRx, an agentic framework that derives RWE for helping with CTD. Through the iterative conversation and analysis across agents with different roles, EmulatRx can autonomously refine trial protocols and finally generate a robust report containing insights that inform better CTD. We applied EmulatRx on the CTD process for both acute diseases (e.g., septic shock, acute heart failure, acute pulmonary edema, and acute kidney injury) using the MIMIC-IV data and chronic diseases (e.g., Alzheimer's disease and Parkinson's disease) using the INSIGHT Network across five New York City health systems. The results demonstrate EmulatRx's capabilities in facilitating and accelerating the CTD process.
9 July 2026
Read appraisal →PloS one
Stylistic analysis of translated languages: A perturbation-based XAI deep learning framework
Text classification using traditional machine learning techniques has been used in natural language processing (NLP) tasks to distinguish translated from non-translated languages, with high accuracy scores indicating the distinctive style of translated languages. While deep learning (DL) has demonstrated impressive performance in terms of representation learning and capturing nuanced patterns in natural language data, DL models act as black boxes, making their results difficult to interpret. This study addresses this issue by demonstrating an explainable AI (XAI) DL framework in a case study of United Nations (UN) meetings. The framework consists of three stages: i) train a variational autoencoder (VAE) combined with BERT embeddings converted from translated and non-translated texts; ii) utilize the majority vote from three classifiers selected from a stacked ensemble to classify the VAE's latent representations; iii) implement a perturbation-based XAI method to interpret the DL model's decisions. The results indicate that the VAE-based model effectively distinguishes the two text types, with accuracy scores above 0.8. The XAI analysis reveals that interpreting the VAE-based model's decision uncovers stylistic differences between the two text types beyond superficial lexical and syntactic features. This proof-of-concept study demonstrates the potential of the XAI DL framework in other NLP studies that aim to analyze style.
9 July 2026
Read appraisal →PloS one
Beyond numbers: The governance role of data assets in real earnings management.
While data assets are increasingly recognized as strategic resources in the digital economy, their implications for financial reporting quality remain underexplored. Employing a textual analysis approach based on deep learning techniques to construct a measure of data assets, this study investigates their impact on real earnings management (REM) using a sample of Chinese A-share listed firms from 2010 to 2023. We document a negative association between data assets and REM. Our findings remain robust after addressing endogeneity concerns using instrumental variables based on regional digital infrastructure and propensity score matching. Mechanism analyses reveal that data assets mitigate REM through two distinct channels: curbing managerial short-termism and strengthening external monitoring via increased analyst coverage. Cross-sectional tests indicate that this governance effect is more pronounced among firms facing earnings pressure. Furthermore, data assets primarily constrain cost-side manipulation strategies, specifically abnormal production costs and discretionary expenditures. Finally, we show that the reduction in REM driven by data assets contributes to enhanced long-term firm value. Collectively, these findings uncover a novel governance effect of data assets, suggesting that digital infrastructure investments yield positive externalities for corporate financial behavior.
9 July 2026
Read appraisal →Journal of medical Internet research
Are Traditional Registries Becoming Obsolete in the Modern Digital Health Ecosystem?
Registries have long been a cornerstone of medical research and public health, providing systematically collected data on diseases, treatments, and health outcomes. However, in the era of digital health, we argue that the traditional model of stand-alone registries needs reconsideration, given the context of increasingly digitized and interoperable health data ecosystems. Unless registries evolve to embrace embedded, standards-based data services, operating across interoperable infrastructure, they will become obsolete while digitalization is reshaping how data can be collected, shared, and used. In this viewpoint, we recount how the present health data ecosystem came to be and what role registries have come to play therein. Following that, we show how recent regulatory initiatives such as the Trusted Exchange Framework and Common Agreement in the United States or the European Health Data Space Regulation signal a shift toward cross-network health information exchange, promoting patient-centric data integration within electronic health record systems. We further illustrate how electronic health records are consequently set to evolve into information hubs, acting as the primary gateway for individuals through which they may access and control their personal health data spread throughout increasingly connected health data ecosystems. This, in turn, might stimulate the creation of digital twins and continuous learning health systems in practice. Following this line of thought, we discuss the opportunities and challenges of interconnected health data ecosystems. Ultimately, we propose that next-generation registries need to be designed as dynamic, service-oriented software stacks for research, leveraging the common data infrastructures that are currently being established around the world. Given the points raised in this viewpoint, we invite health care professionals and researchers alike to equally rethink the role that registries should play within the globally emerging interconnected health data ecosystems and contribute their findings. References included in this viewpoint were identified through searches of PubMed and Google Scholar with various search terms and combinations thereof pertinent to the topics touched on, for example, "patient registry," "clinical registry," "digital twin," "healthcare," "clinical research," "virtual twin," "TEFCA," or "EHDS." Only papers in English were reviewed. The final reference list was generated on the basis of originality and relevance to the broad scope of topics covered in this viewpoint, aiming to present a balanced overview of topic-related findings and arguments.
5 July 2026
Read appraisal →JMIR formative research
Knowledge, Attitudes, and Practices Related to AI in Learning and Research Among Medical Students in Vietnam: Cross-Sectional Study
BACKGROUND: In recent years, artificial intelligence (AI) has ushered in a promising era in medicine, particularly in medical education. However, studies assessing the knowledge, attitudes, and practices related to AI among medical students in Vietnam remain limited. OBJECTIVE: This study aimed to evaluate AI knowledge, attitudes, and practices among Vietnamese medical students in learning and research, and to identify factors associated with their AI practices. METHODS: A cross-sectional study was conducted among medical students at Thai Binh University of Medicine and Pharmacy from November to December 2025. Data were collected using an online structured questionnaire covering demographic characteristics and AI knowledge, attitudes, and practices. The main outcome of interest was AI practices in learning and research. Descriptive statistics and multivariable linear regression were used to examine associated factors. Regression coefficients (β), 95% CIs, and P values are reported. RESULTS: A total of 1002 medical students (mean age 21.00, IQR 19.00-23.00 years; n=596, 59.5% female) were included. The median percentage of maximum possible (POMP) score of AI knowledge was 66.67 (IQR 33.33-83.33), with a high level of familiarity with common tools (n=798, 79.6%). AI attitudes were generally positive (median POMP score 70.00, IQR 53.33-76.67). AI-related practices were lower (median POMP score 50.00, IQR 46.88-71.88), with AI being used primarily for information retrieval and literature research support. In the multivariable analysis, knowledge POMP score (β=0.12, 95% CI 0.08-0.16) and attitudes POMP score (β=0.42, 95% CI 0.34-0.51) were significantly associated with AI practices POMP score (P<.001). Age, gender, major, grade point average classification, and having participated in an AI seminar or training were not associated with AI practices. CONCLUSIONS: Medical students showed favorable knowledge and positive attitudes, but their AI practices remained limited. Integrating AI into medical curricula, including fundamentals, applications, and ethical aspects, is essential to prepare future physicians for AI-driven health care.
5 July 2026
Read appraisal →Computers, informatics, nursing : CIN
Bridging the Gap Between Potential and Practice: An Integrative Review of Generative Artificial Intelligence in Nursing
BACKGROUND: Generative artificial intelligence (GAI) has emerged as a transformative tool in health care, particularly with the integration of large language models (LLMs) into electronic health record (EHR) systems. These tools have the potential to enhance clinical decision-making, increase efficiencies, and decrease time in the EHR. However, despite expanding availability, adoption in nursing practice remains limited. This integrative review aimed to synthesize the literature and explore barriers related to the integration of GAI tools into nursing practice. METHODS: We conducted an integrative search of peer-reviewed literature published between November 2022 and July 2025 to examine the current evidence while assessing for methodological quality. The retrieved articles were screened for title, abstract, and full text eligibility. We synthesized themes related to GAI adoption by nurses, focusing on ethical, educational, and workflow-related factors that influence use. RESULTS: A total of 10 studies met criteria including 7 qualitative descriptive design, 2 case reports, and 1 experimental clinical trial. Four key barriers emerged: educational gaps, ethical concerns, data transparency issues, and workflow misalignment contributing to nurses' hesitancy to engage with GAI tools. Recommendations include involvement of nurses in the design, implementation, and evaluation of GAI tools and mandatory AI curricula in nursing education. CONCLUSIONS: Overcoming barriers requires nurse involvement in GAI design efforts, targeted education, and governance models that foster trust and usability. Nurse-centered integration of GAI tools has the potential to advance workforce efficiencies while preserving patient safety.
4 July 2026
Read appraisal →JMIR mental health
From Alliance to Nexus: Rethinking Digital Therapeutic Relationships
In traditional human psychotherapy, the therapeutic alliance (TA) is regarded as a fundamental factor that describes the client-therapist relationship, mainly due to strong evidence demonstrating its impact on treatment outcomes regardless of theoretical orientation. More recently, advances in artificial intelligence (AI) and other technologies have led to the emergence of the concept of digital TA, used to characterize the relationship between clients and AI-based therapeutic systems. This approach replicates human dynamics but overlooks key differences between human therapists and digital agents. Prematurely translating the concept of TA into the digital context fails to address issues such as the sycophantic tendencies of current systems and the inherent limitations of algorithmic interaction. We propose the digital therapeutic nexus, a framework that recognizes these differences and provides a set of structured criteria for categorizing digital interactions into 3 progressive levels. This Viewpoint argues that only at the highest level can parallels be drawn to the human TA and stratifies the main risks associated with each nexus level. Transitioning from the concept of alliance to that of a nexus offers a more precise conceptual basis for describing and evaluating digital therapeutic relationships, with implications for research, design, and the ethical development of AI-based mental health interventions.
3 July 2026
Read appraisal →Korean journal of radiology
Agentic Artificial Intelligence for the Automated Generation of Accurate Summary Podcasts of Radiology Research Papers
OBJECTIVE: To evaluate whether a custom agentic artificial intelligence (AI) pipeline can overcome the limitations of general-purpose large language model tools, when compared with a generic commercial tool (Google NotebookLM [NBLM]), for generating podcast-style summaries of radiology research articles. MATERIALS AND METHODS: Twenty-two PDF-format original research articles published in the April 2025 issue of Radiology were processed using our Programmable, Phoneme-Aware PDF-to-Podcast Pipeline (P5) and NBLM to generate 44 audio episodes. P5 utilizes a multi-agent workflow for script generation, quality assurance, pronunciation enhancement, and audio synthesis. Four radiologists from a pool of 25 (7 generalists and 18 specialists) were randomly assigned to evaluate each blinded audio episode, yielding 176 total evaluations. The primary outcomes were the number of hallucinations (factual errors) per episode and the percentage of hallucination-free episodes. Secondary outcomes included the number of inappropriate statements, mispronunciations, and flow disruptions; the composite quality score (Quality Assessment of Educational Podcasts [QAEP]); the key results coverage score; and overall listener preference. Data were analyzed using generalized linear mixed models. RESULTS: The P5 method produced significantly fewer hallucinations per episode compared with NBLM (mean, 0.32 vs. 0.93; P < 0.001) and a higher proportion of hallucination-free episodes (71.6% [63/88] vs. 56.8% [50/88]; P = 0.013), consistently across generalists and specialists. P5 demonstrated significantly fewer mispronunciations (mean, 0.11 vs. 1.62; P < 0.001) and flow disruptions (mean, 0.19 vs. 1.06; P < 0.001) per episode, as well as higher mean QAEP composite scores (4.62 vs. 4.28; P < 0.001), compared with NBLM, consistently across generalists and specialists. Raters preferred P5 over NBLM in 72.7% (64/88) of comparisons (P = 0.003). CONCLUSION: Our custom agentic AI pipeline generated podcast-style summaries of radiology research articles with significantly higher quality and greater listener preference than the generic commercial tool.
3 July 2026
Read appraisal →Nature communications
A deep learning framework for efficient pathology image analysis
Artificial intelligence has transformed digital pathology by enabling biomarker prediction from high-resolution whole-slide images. However, current methods are computationally inefficient, processing thousands of redundant tiles per slide and requiring complex aggregation models. We introduce EAGLE (Efficient Approach for Guided Local Examination), a deep learning framework that emulates pathologists by selectively analyzing informative regions. EAGLE combines task-agnostic tile selection with detailed feature extraction and is benchmarked against leading slide- and tile-level foundation models across 43 tasks from nine cancer types spanning morphology, biomarker prediction, treatment response and prognosis. EAGLE outperforms patch aggregation methods by up to 23% and achieves the highest overall classification performance. It processes one slide in 2.27 s, reducing computational time by more than 99% compared with existing models. This efficiency supports rapid and auditable workflows by enabling review of the exact tiles used for each prediction and reducing dependence on high-performance computing. By reliably identifying informative regions and minimizing artifacts, EAGLE provides robust and auditable outputs, supported by systematic negative controls and attention concentration analyses. Its unified embedding enables rapid slide search, integration into multi-omics pipelines and emerging clinical foundation models.
2 July 2026
Read appraisal →Journal of the American Medical Informatics Association : JAMIA
Clinicians' rationale for editing ambient AI-drafted clinical notes: persistent challenges and implications for improvement
OBJECTIVE: The use of ambient AI documentation tools is rapidly growing in US hospitals and clinics. Such tools generate the first draft of clinical notes from scribed patient-provider conversations, which clinicians can then review and edit before signing into electronic health records (EHR). Understanding how and why clinicians make modifications to AI-generated drafts is critical to improving AI design and clinical efficiency, yet it has been under-studied. This study aims to address this gap. MATERIALS AND METHODS: We conducted semistructured interviews with 30 clinicians from the University of California, Irvine Health who used a commercial ambient AI tool in routine outpatient care. We invited them to describe how and why they edited AI drafts based on both their personal experience and review of some real-world examples identified from our previous studies. RESULTS: Modifications to AI drafts were primarily made to improve clinical accuracy and specialty-specific precision, reduce medico-legal and liability risk, and meet billing, coding, and documentation standards. Such editing was necessary due to reasons such as transcription errors, speaker attribution mistakes, overconfident statements without evidence, missing key clinical details, and AI's lack of information about the patient context. CONCLUSION AND DISCUSSION: Improving ambient AI documentation will require coordinated effort from vendors, institutions, and clinicians. Key targets include core model reliability (eg, transcription accuracy), specialty- and encounter-level customization, clinician-level personalization, more effective EHR integration, and institutional support (eg, training, governance, and standardized review guidance), complemented by clinicians' adaptive communication strategies that strengthen human-AI collaboration.
2 July 2026
Read appraisal →PloS one
Developing a natural language processing system using transformer-based models for adverse drug event detection in electronic health records
OBJECTIVE: To develop a transformer-based natural language processing (NLP) system for detecting adverse drug events (ADEs) from clinical notes in electronic health records (EHRs). MATERIALS AND METHODS: We fine-tuned BERT Short-Formers and Clinical-Longformer using the processed dataset from the 2018 National NLP Clinical Challenges (n2c2) shared task Track 2. Two data processing methods, window-based and split-based approaches, were compared to identify the optimal processing method. Model generalizability was evaluated on a dataset extracted from Vanderbilt University Medical Center (VUMC) EHRs. RESULTS: On the n2c2 dataset, the best 5-fold cross-validation AUPRC, micro F1, and macro F1 scores were 0.840 (Clinical-Longformer, 4-chunk split), 0.965 (BioBERT, 15-word window), and 0.852 (Clinical-Longformer, 10-chunk split). On the VUMC dataset, the best AUPRC, micro F1, and macro F1 scores were 0.536 (Clinical-Longformer, 6-chunk split), 0.966 (BERT-base-uncased, 6-chunk split), and 0.762 (Clinical-Longformer, 4-chunk split). DISCUSSION: Transformer-based models demonstrated strong performance for ADE detection, with split-based processing generally outperforming window-based methods. Clinical-Longformer combined with a practical split-based approach showed promise for real-world implementation. Beyond token limits, chunk size substantially influenced model performance, even when text length remained within limits. CONCLUSION: Our findings provide practical guidance for developing transformer-based ADE detection systems from clinical notes. The selection of both text preprocessing strategies and model architectures should be guided by note characteristics and practical considerations such as annotation burden.
2 July 2026
Read appraisal →The Behavioral and brain sciences
Large language models have learned to use language.
Acknowledging that large language models have learned to use language can open doors to breakthrough language science. Achieving these breakthroughs may require abandoning some long-held ideas about how language knowledge is evaluated and reckoning with the difficult fact that we have entered a post-Turing test era.
2 July 2026
Read appraisal →The Behavioral and brain sciences
Large language models are not about natural language.
Large Language Models are useless for linguistics, as they are probabilistic models that require a vast amount of data to analyze externalized strings of words. In contrast, human language is underpinned by a mind-internal computational system that recursively generates hierarchical thought structures. The language system grows with minimal external input and can readily distinguish between real language and impossible languages.
2 July 2026
Read appraisal →IEEE transactions on pattern analysis and machine intelligence
50 Years of Automated Face Recognition
Over the past five decades, automated face recognition (FR) has progressed from handcrafted geometric and statistical approaches to advanced deep learning architectures that now approach, and in many cases exceed, human performance. This paper traces the historical and technological evolution of FR, encompassing early algorithmic paradigms through to contemporary neural systems trained on extensive real and synthetically generated datasets. We examine pivotal innovations that have driven this progression, including advances in dataset construction, loss function formulation, network architecture design, and feature fusion strategies. Furthermore, we analyze the relationship between data scale, diversity, and model generalization, highlighting how dataset expansion correlates with benchmark performance gains. Recent systems have achieved near-perfect large-scale identification accuracy, with the leading algorithm in the latest NIST FRTE 1:N benchmark reporting a False Negative Identification Rate (FNIR) of 0.15 percent at False Positive Identification Rate (FPIR) of 0.001 on a gallery of over 10 million identities. Larger galleries increase false positive rates and deployments at greater scales will see higher error rates. We delineate key open problems and emerging directions, including scalable training, multi-modal fusion, synthetic data, and interpretable recognition frameworks.
1 July 2026
Read appraisal →Journal of integrative medicine
Deep learning in traditional Chinese medicine
As an alternative medicine approach, traditional Chinese medicine (TCM) is mainly built on an ancient understanding of the world. In order to translate the promises of TCM into modern therapies, rigorous scientific techniques are applied, which generates big data from experiments, clinical practices and medical literature. Recently, deep learning (DL) has attracted the attention of TCM researchers due to its application in data mining. In this review, we summarized recent examples of DL application in the study of TCM, including medical image processing, medicinal substances investigation, data fusion and natural language processing. We find that DL methods have considerable potential to extract valuable information from TCM data, but high-quality TCM data are still needed for further development of DL techniques. Please cite this article as: Liu LP, Yang C, Cen SX, Suo TC. Deep learning in traditional Chinese medicine. J Integr Med. 2026; 24(4):471-480.
1 July 2026
Read appraisal →Journal of the American College of Clinical Pharmacy : JACCP
Research and Scholarly Methods: Using Artificial Intelligence
Artificial intelligence (AI) is quickly changing health care and expanding research opportunities, including in pharmacy. AI can offer effective solutions for numerous challenges in pharmacy research, such as persistent gaps in evidence-based prescribing, slow drug discovery-to-adoption timelines, complex data environments, and the need for more efficient methods, including qualitative methods. AI tools-ranging from machine learning models to large language models (LLMs)-offer new capabilities for data analysis, prediction, and content generation. Pharmacy researchers can harness these technologies to address persistent gaps in medication management, optimize patient safety and outcomes, and streamline many research workflows. However, successful integration requires understanding the spectrum of AI capabilities, appropriate use cases, and potential limitations. AI must be tailored to the research question, data, and institutional resources. This paper will guide pharmacy researchers in identifying use cases for AI in their research, selecting and applying AI tools, mitigating risks, providing standards for reporting the use of AI in research, and maximizing benefits for scientific discovery and patient outcomes.
1 July 2026
Read appraisal →Medical education online
Exploring medical Artificial Intelligence Readiness among Korean medical students: a cross-sectional study
INTRODUCTION: Artificial intelligence (AI) is rapidly transforming healthcare systems and clinical practice, increasing the need to prepare future physicians to work effectively with AI technologies in clinical contexts. Despite growing interest in integrating AI into medical education, empirical evidence regarding medical students' readiness to use AI remains limited. This study aimed to assess medical AI readiness among Korean medical students and explore factors associated with their readiness. METHODS: A cross-sectional survey was conducted among medical students enrolled in a six-year medical program at a university in Korea. A total of 204 students participated in the study. Medical AI readiness was measured using the Medical Artificial Intelligence Readiness Scale for Medical Students (MAIRS-MS), which assesses four domains: cognition, ability, vision, and ethics. Descriptive statistics, independent t-tests, and Pearson's correlation analyses were performed using SPSS version 27. RESULTS: The overall mean score of medical AI readiness was 4.19 on a 7-point Likert scale, indicating a moderate level of medical AI readiness. Among the subscales, ethics showed the highest mean score (4.69), followed by vision (4.44), ability (4.18), and cognition (3.92). The frequency of AI use was significantly associated with medical AI readiness, whereas the daily duration of AI use was not. No significant gender differences were observed. When students were divided into the low (pre-medical years 1-2) and high (medical years 1-4) groups, the high group showed significantly higher scores only in the ability subscale. CONCLUSIONS: The findings suggest that medical students are not yet sufficiently prepared to utilize AI technologies in clinical practice, particularly in terms of knowledge and practical competencies related to AI. These results highlight the need for structured and longitudinal AI education in medical curricula to better prepare future physicians for AI-integrated healthcare environments.
1 July 2026
Read appraisal →PloS one
An art style classification network integrating contrastive learning and counterfactual attention
Art style classification is a fundamental task for digital art analysis and intelligent cultural heritage management. However, existing methods often fail to deliver stable and interpretable results due to dispersed style cues, complex local textures, strong interference from background and subject content, and the difficulty of identifying regions that are causally relevant to style recognition. To address these issues, we propose a Multi Source Collaborative Style Network, termed MCS-Net, which jointly models global composition and local brushstroke textures within a unified framework. MCS-Net consists of four modules, namely a basic style feature encoding module, an attention generation module, a style contrastive learning module, and a counterfactual attention module. The attention generation module discovers multiple complementary discriminative regions. The contrastive learning module constructs a more discriminative embedding space to improve the separability of fine-grained styles. The counterfactual attention module builds counterfactual attention samples to explicitly estimate the true contribution of regions to predictions, thereby suppressing spurious activations, improving robustness, and providing counterfactual-supported evidence for model interpretation. Experiments on three public datasets, WikiArt, MultitaskPainting100k, and Pandora18k, demonstrate that MCS-Net outperforms representative baselines across standard evaluation metrics.
28 June 2026
Read appraisal →Journal of health organization and management
Cloud-based healthcare architecture for securing and monitoring healthcare data
PURPOSE: A new method known as Lionized Remora optimization based Recurrent Neural Network (LRObRNN) is recommended to enhance the safety of medical information stored on cloud servers to tackle these issues. DESIGN/METHODOLOGY/APPROACH: To safeguard patient data, healthcare organizations must thoughtfully choose reliable and compliant cloud service providers while implementing robust security measures. Storing patient information in cloud systems raises issues with illegal access and data breaches. FINDINGS: The LRObRNN generates a secret key using Lionized Remora optimization and employs cryptography to encrypt sensitive healthcare data. Continuous monitoring ensures the security of data transmission by identifying irregularities. ORIGINALITY/VALUE: Leveraging Recurrent Neural Networks the system analyzes sequential data, enabling the detection of patterns and potential security breaches during data transmission. The performance evaluation includes metrics such as encryption and decryption time, confidentiality rate, processing time, resource usage and efficiency, which are compared with other existing models.
27 June 2026
Read appraisal →PloS one
Transformer-based advertisement image layout generation and object fidelity optimization
The layout of the image and the positional distribution of the objects directly affect the audience's visual focus and the effect of the message conveyed. By setting up a specific layout structure, the viewer's attention can be more focused on the product launched or the moral of the advertisement. Existing work related to position-controllable text-to-image generation has made great progress in generating results on simple images. However, when generating images of complex scenes, the quality is often poor. This can result in the model failing to accurately convey the message of the advertisement when used to generate advertisements. To address these limitations, we propose LAYOBJ-GAN, a novel two-stage framework for layout-controllable advertisement image synthesis. Unlike prior work, our method explicitly models background layouts jointly with object layouts during the text-to-layout generation stage, enabling comprehensive spatial planning of complex scenes. Technically, we introduce a Transformer-based sequence-to-sequence layout generator that learns long-range dependencies between textual descriptions and both background and object regions, which has not been explored in previous advertisement-oriented text-to-image frameworks. In the layout-to-image stage, we further propose a fine-grained text-layout interaction normalization module (TL-Norm) that enables object knowledge transfer from a pre-trained category-to-image model, allowing object appearance to be adaptively modulated by textual context and layout constraints. Extensive experiments on MS-COCO and a high-definition advertisement dataset (AsHQ-10K) demonstrate that LAYOBJ-GAN significantly outperforms seven state-of-the-art methods in image quality, layout controllability, and semantic object accuracy. These results confirm the effectiveness of explicitly modeling background layouts and transferring object-level generative knowledge for complex advertisement image synthesis.
27 June 2026
Read appraisal →JMIR medical education
Exploring the Role of Early Career Medical Professionals From a Digital-Oriented University in Germany in Promoting Digital Health in Professional Settings: Qualitative Interview Study
BACKGROUND: To address care delivery gaps, the health care system must embrace innovative digital solutions. Additionally, the rising integration of digitalization as a topic into medical education is providing students with broader opportunities to engage with digitalization overall. As digital health becomes an increasingly integral component of medical education and health care practice, digitally affine early career medical professionals constitute a vital resource for advancing digitalization within the health care sector. OBJECTIVE: This study examines how early career medical professionals from Witten/Herdecke University in Germany, with various courses focusing on digitalization, acquire digital knowledge, apply it across diverse practical contexts-ranging from start-ups and corporate environments to traditional clinical settings-and how they contribute to the advancement of digitalization within the entire health care sector. METHODS: Using a qualitative approach, 19 interviews were conducted with early career medical professionals who graduated in the last 15 years at Witten/Herdecke University. Subsequently, the interviews were transcribed and analyzed using a deductive-inductive approach. RESULTS: The findings reveal that medical graduates often acquire digital skills through intensive self-study and learning on the job, integrating them in various ways into their professional lives. Moreover, while graduates recognize their high potential to make their own contribution to the advancement of digitalization, they also face significant barriers such as knowledge gaps, limited resources, and complex regulations, which hinder their ability to contribute to digitalization in a variety of professional settings. Medical graduates report that they face a pressing need for enhanced knowledge access, improved institutional frameworks, and supportive policy measures to maximize their potential in advancing digitalization initiatives. CONCLUSIONS: Recent medical graduates represent an underused resource for health care digitalization. Unlocking this potential requires coordinated action across medical education, health care institutions, and policymaking to create appropriate conditions for graduates to actively drive digitalization.
26 June 2026
Read appraisal →European journal of hospital pharmacy : science and practice
A bibliometric analysis of the application trends of information technology in antimicrobial stewardship within hospitals
OBJECTIVES: This study uses bibliometric analysis to systematically map research trends, knowledge structure and evolution of information technology (IT) in hospital antimicrobial stewardship (AMS) over the last 20 years. METHODS: A Web of Science Core Collection search (2000-2025) yielded 258 English-language publications on IT applications in hospital AMS. A bibliometric analysis, utilising CiteSpace and Bibliometrix, quantitatively evaluated domain evolution through temporal analysis, network mapping, keyword clustering, burst detection and examination of highly cited publications. RESULTS: The bibliometric analysis reveals a fluctuating yet overall increasing trend in annual publications, peaking at 49 articles in 2024. The US (131 publications) and European nations demonstrate significant research output (centrality >0.2), with major collaborative networks coalescing around institutions including Harvard University and Imperial College London. Keyword analysis identifies 'AMS' as a core theme, closely associated with technological keywords such as 'machine learning (ML)', 'clinical decision support systems (CDSS)', 'electronic health records (EHR)' and 'artificial intelligence (AI)'. Emerging trends suggest a shift in research focus from foundational strategies to data-driven prediction of antimicrobial resistance (AMR) and precision interventions. Highly cited literature emphasises the integration of EHR and ML technologies for optimising prescriptions and predicting resistance patterns. CONCLUSIONS: IT-driven AMS research has shifted from empirical management to data science. Despite EHR integration, and ML and CDSS support, challenges remain in data standardisation, technical deployment and ethics. Future work must emphasise global collaboration, standardisation, design refinement and ethical guidelines, and provide clear algorithm explanations to enhance AMR mitigation.
26 June 2026
Read appraisal →Journal of medical Internet research
You Can't Launch This: Trust as Infrastructure in Digital Behavioral Health
Digital behavioral health interventions frequently fail to scale, even when evidence-based and technically and operationally sound. In this News and Perspectives article, researcher, digital behavioral health platform founder, and JMIR Correspondent Trevor van Mierlo concludes a four-part series examining why this occurs, reporting on the foundational role of trust.
25 June 2026
Read appraisal →PloS one
A lightweight hybrid deep learning approach for fashion mnist classification with explainable attention visualization
The classification of fashion images is an essential task in the e-commerce sector, where accurate categorization improves user experience and refines product discovery. Convolutional Neural Networks (CNNs) and Transformers have demonstrated strong performance in image classification tasks due to their ability to learn complex visual features. However, deep variants of these architectures, such as VGG-19, ResNet-50, Vision Transformer (ViT), and Swin Transformer, contain tens of millions of parameters, requiring high memory and powerful GPUs for training, which makes them less suitable for low-resource and edge device environments. To address these limitations, this research proposes a lightweight hybrid architecture, TinyCNN with Linear Self-Attention (LSA), optimized for resource-constrained settings. The proposed model contains fewer than half a million parameters and is trainable on a CPU, achieving a classification accuracy of 91.47% on the Fashion-MNIST dataset. In addition, multiple Explainable Artificial Intelligence (XAI) techniques are implemented, including Self-Attention visualization, Multi-Head Attention, Attention Flow, Attention Rollout, Fixed query position attention maps, Integrated Gradients, LIME, and SHAP, to provide visual interpretability of the model's predictions and enhance transparency in its decision-making process.
25 June 2026
Read appraisal →JMIR nursing
Digital Health Literacy and Attitudes Toward Telehealth Use in Practice Among Nursing Students in Saudi Arabia: Cross-Sectional Study
BACKGROUND: Nursing students are the future workforce, and their readiness to use digital health tools is important. Previous studies have focused on knowledge and attitudes; however, they have not examined the wide range of digital health literacy levels that may influence nursing students' attitudes toward using telehealth in clinical settings. OBJECTIVE: This study aimed to determine the relationship between nursing students' digital health literacy and their attitudes toward telehealth use in practice. METHODS: A cross-sectional design was used. The sample consisted of undergraduate nursing students enrolled in a Bachelor of Nursing program at a selected Saudi Arabian university. The online survey used 2 scales: the Digital Health Care Literacy Scale and the Nurses' Attitudes Toward Use of a Telehealth Scale. RESULTS: A total of 273 students participated (mean age 21.3, SD 1.9 years). Most of the nursing students demonstrated a high digital health literacy level (n=184, 67.4%; mean Digital Health Care Literacy Scale score 11.9 out of 15, SD 3.1). Digital health literacy was a significant predictor of positive attitudes toward telehealth use in practice (adjusted odds ratio 1.48, 95% CI 1.28-1.71; P<.001). Male students were significantly less likely to report positive attitudes than female students (adjusted odds ratio 0.62, 95% CI 0.39-0.97; P=.03). However, academic year, telehealth workshops, and informatics courses were not significantly associated with positive attitudes toward telehealth use in practice. CONCLUSIONS: Higher levels of literacy appear to correlate with more positive attitudes toward telehealth use in practice. However, current formal education and workshops had no apparent influence on digital health literacy. This suggests a potential need for strengthening digital training and development in nursing education. This may enhance telehealth readiness and support future digital health care delivery.
24 June 2026
Read appraisal →BMJ health & care informatics
Clarifying the relationship between biomedical and health informatics and digital health: expert perspectives.
The rise of digital health (DH) has sparked debates about its relationship with the long-established field of biomedical and health informatics (BMHI), with some questioning the need for new academic programmes in DH since much of the likely content is already covered in BMHI training programmes. At the International Medical Informatics Association panel, 'Shaping the Future of Biomedical and Health Informatics in the Digital Health Era' (MEDINFO 2025), only one-third of participants clearly distinguished the two fields, highlighting conceptual ambiguities and the influence of DH narratives. This article compiles the perspectives expressed by the panellists, all experienced BMHI scholars, presenting their distinct viewpoints on how the field can maintain its scientific identity while engaging with emerging DH initiatives.
24 June 2026
Read appraisal →JMIR medical informatics
Blockchain-Based Dynamic and Revocable Consent for Secondary Health Data Use: Systematic Review
BACKGROUND: The secondary use of health data holds substantial potential for advancing biomedical research, strengthening population health analytics, and enabling artificial intelligence-driven decision-making support. Yet, ensuring that such reuse respects patient autonomy, privacy, and regulatory obligations remains a major challenge. Conventional consent mechanisms are typically static, difficult to revoke, and offer limited transparency or accountability after data disclosure. OBJECTIVE: This review aimed to systematically examine blockchain-based frameworks that enable dynamic, auditable, and revocable consent for the secondary use of health data. METHODS: A structured literature search was conducted in PubMed, Scopus, and Web of Science covering the period 2020 to 2025. Following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, 55 peer-reviewed studies meeting predefined inclusion criteria were analyzed. Data extraction focused on four dimensions: (1) consent life cycle management, (2) auditability and traceability, (3) usability and patient empowerment, and (4) legal and ethical alignment. RESULTS: Findings indicate that blockchain technologies provide a robust foundation for automating consent life cycles, ensuring immutable auditability, and enabling decentralized patient control. Most frameworks used smart contracts, decentralized identifiers, and verifiable credentials to implement programmable and verifiable consent processes. Nevertheless, key challenges persist, including limited usability testing, complexities in real-time revocation propagation, interoperability gaps with clinical systems, and tensions with regulatory requirements such as the General Data Protection Regulation right to erasure. Only a small subset of studies reported real-world deployments or user-centered evaluations. CONCLUSIONS: Blockchain offers substantial promise for improving the trustworthiness, transparency, and accountability of consent management for secondary health data use. However, wider adoption requires human-centered design approaches, stronger interoperability through standards such as Fast Healthcare Interoperability Resources, verifiable credentials, and consent receipts, and clearer legal guidance for compliance. Future research should prioritize integrating blockchain-enabled consent infrastructures into national and cross-border digital health ecosystems such as the European Health Data Space to support secure, patient-controlled, and ethically governed secondary data use.
24 June 2026
Read appraisal →Archives of disease in childhood
AI-simulated clinical consultations: Assessing the potential of ChatGPT to support medical training
BACKGROUND: Simulated medical scenarios are useful for evaluating and developing clinical competencies but scheduling them is expensive and time-consuming. Large language models show promise in role-playing tasks. We investigated the fidelity with which ChatGPT can mimic patients, clinicians and examiners in educational settings. OBJECTIVE: To determine the realism with which ChatGPT can portray patient, doctor and examiner roles, and the utility of these agents in clinical education. METHOD: We selected four paediatric scenarios from mock objective structured clinical examinations (OSCEs) and set up separate patient, doctor and examiner ChatGPT agents for each. The patient and doctor agents conversed with each other in written format. The examiner agent marked the doctor agent based on this conversation. Patients and clinicians familiar with the OSCE assessed the dialogues. RESULTS: The patient agent was judged to be true to character most of the time and good at expressing emotion. The doctor agent was reported to be an effective communicator but occasionally used jargon. Both agents tended to produce repetitive responses which undermined realism. The examiner agent had good correlation with human clinicians. There was moderate support for using the simulated interactions for educational purposes. CONCLUSION: Although the realism of the agents can be improved, ChatGPT can generate plausible proxies of participants in medical scenarios and could be useful for complementing standardised patient-based training.
24 June 2026
Read appraisal →BMJ open
Development of phenotype algorithms for the detection of adverse events in electronic health record data: a multicentre study
OBJECTIVES: To develop phenotype algorithms for the detection of adverse events (AEs) or AE-proxies in electronic health records (EHRs), accounting for varying data availability. DESIGN: Multicentre study conducted as part of the Use Case POLAR_MI (POLypharmacy, drug interActions, Risks) of the German Medical Informatics Initiative (MII). SETTING: Germany. PARTICIPANTS: Multidisciplinary teams from 10 German university sites within the MII. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Literature- and consensus-based development and operationalisation of AE algorithms using structured EHR data, including a standardised, multicentre expert review process. Data categories used: International Classification of Diseases, 10th Revision (ICD-10) codes for diagnoses; Anatomical Therapeutic Chemical (ATC) codes and 'Pharmazentralnummern' (PZN; German eight-digit identification code for pharmaceutical products) for medications (used in the treatment of AEs); Logical Observation Identifiers Names and Codes (LOINC) for laboratory values and medical findings; and 'Operationen- und Prozedurenschlüssel' (OPS; German procedure classification) codes for medical and surgical procedures. RESULTS: We developed 82 algorithms for 48 AEs. Algorithms for the same AE varied by data categories or code selections. At the AE level, 31 AEs were covered exclusively by newly developed algorithms, and 17 AEs by at least one modified algorithm.Overall, 52 algorithms were based on a single data category, while 30 required multiple categories. ICD-10 codes were most commonly used (n=65 AE algorithms), followed by LOINC (n=27), ATC codes (n=18), OPS codes (n=11) and PZN (n=2). All phenotype algorithms were semantically modelled and can be executed using the publicly available Terminology- and Ontology-based Phenotyping (TOP) Framework, which supports export in various formats. CONCLUSION: We present a peer-reviewed set of algorithms for a large number of AEs, which can be implemented in structured routine electronic data sources and (pending validation studies) may support pharmacoepidemiologic research. The algorithms will be implementable across all 39 participating sites of the German MII. As a next step, we will empirically validate the algorithms against all information (including free text) contained in EHRs. TRIAL REGISTRATION: Not applicable.
24 June 2026
Read appraisal →JMIR nursing
Digital Leadership Scale for Clinical Nurses: Development and Validation of an Instrument
BACKGROUND: The rapid advancement of digital technologies, combined with the evolving complexity of health care environments, has introduced a new paradigm in nursing practice. Clinical nurses are now required not only to deliver safe and effective patient care but also to demonstrate competencies in digital literacy and innovation. Among these emerging competencies, digital leadership has become a critical attribute-enabling nurses to lead digital transformation, ensure patient safety, enhance care quality, and support system-level change within health care organizations. Despite its increasing relevance, there is a notable absence of validated measurement tools tailored to assess digital leadership in clinical practice. OBJECTIVE: This study aimed to develop and psychometrically validate a Digital Leadership Scale for Clinical Nurses (DLS-CN) to systematically evaluate the digital leadership capabilities of nurses working in clinical settings. METHODS: The scale development process followed a rigorous multistep procedure. Initial items were derived from previous qualitative research involving a literature review and in-depth interviews, complemented by an additional literature review conducted in this study. The content validity of 38 preliminary items was evaluated by 9 experts over 2 rounds. A pilot test was conducted with 30 nurses, followed by cognitive interviews with 5 nurses to refine item clarity and relevance. The final set of items was administered to 446 clinical nurses across various health care institutions. Data were randomly split for exploratory factor analysis and confirmatory factor analysis. Additional analyses were conducted to evaluate item discrimination, convergent validity, and internal consistency using IBM SPSS 25.0 and AMOS 23.0. RESULTS: The finalized DLS-CN consists of 29 items grouped under four domains: (1) ability to use digital technology, (2) digital safety management, (3) digital collaboration mindset, and (4) organizational influence. These 4 factors explained 56.9% of the total variance. The scale showed strong internal consistency (Cronbach α=0.95). Convergent validity was demonstrated through strong positive correlations with the Nursing Informatics Competency Scale (Pearson correlation coefficient r=0.82; P<.001) and the Self-Leadership Scale (Pearson correlation coefficient r=0.83; P<.001). CONCLUSIONS: The DLS-CN is a valid and reliable instrument for measuring digital leadership among clinical nurses. It offers a practical tool for educators, administrators, and researchers to assess and enhance digital leadership capabilities-ultimately supporting the digital transformation of health care systems.
24 June 2026
Read appraisal →Journal of psycholinguistic research
Statistical Analysis of Human and AI Machine Translation Accuracy, Style, and Context in Academic English
AI-driven Machine Translation (MT) technology has raised doubts about whether it can replace Human Translation (HT), particularly in the academic English field. This research contrasts AI-based MT and HT of academic English on accuracy, stylistic quality, errors, contextual appropriateness, and overall suitability. A quantitative comparative design was employed for the translation of eight Chinese texts with various genres. Using purposive sampling, 4 certified human translators were selected to produce the HT, and also used AI-based MT systems such as DeepL, ChatGPT, and Google Translate to translate the text. 69 experts evaluated the translated versions, and this data was examined using SPSS software. HT produced better performance than MT through mean accuracy, stylistic quality, contextual appropriateness, and overall suitability (Human: 4.76, AI: 3.73), revealed statistically significant differences (p < 0.001), and effect sizes (Cohen's d = 1.53 to 2.81). Compared to HT, the MT showed higher grammatical and semantic errors (Cohen's d = 2.81). The fluency, tone, and register of stylistic quality were evaluated at higher levels for HT (p < 0.001). Among AI-MT tools, DeepL shows higher performance in translation (M = 4.16, SD = 0.37). The genre-wide comparison of MT and professionally certified HT, and the comparison among multiple AI tools, towards various translation dimensions, provides insights into the differences in translating quality. The research indicates a continued reliance on human professionals for important scholarly translations while encouraging continued development in AI translation software for better quality.
22 June 2026
Read appraisal →JMIR cancer
Digital Intervention for Electronic Patient-Reported Outcomes in Advanced Cancer: Mixed Methods Study
BACKGROUND: Digital health interventions are increasingly being integrated into oncological care to support patients in managing treatment-related symptoms and psychological distress. In a randomized controlled pilot trial, we investigated the feasibility and preliminary efficacy of a digital health app (SOFIA) among patients with cancer, including those in palliative care. SOFIA consists of an electronic patient-reported outcome (ePRO) assessment and coaching component. We showed good feasibility and high acceptability of SOFIA in routine clinical care. OBJECTIVE: This pilot study aimed to evaluate user experiences and behavior with SOFIA. We applied a mixed methods design, combining a qualitative exploration of patients' experiences and a quantitative analysis of app usage patterns. The integrated findings are intended to inform the refinement and further development of digital health interventions to better address the needs, preferences, and engagement behaviors of this patient population. METHODS: Patients randomly assigned to the intervention group participated in semistructured interviews at 2 time points during the 12-week intervention period: midway through the intervention (at week 6) (T1) and postintervention (at week 12) (T2). Qualitative data were analyzed using content analysis. User data were descriptively analyzed with Python and the Pandas library (version 2.2.3). RESULTS: Our qualitative analysis of 29 patients revealed benefits regarding the ePRO assessment (empowerment, support, user-friendliness, and facilitation of patient-physician communication), as well as some criticism (inflexible design and insufficient use by physicians). Benefits of the coaching tool included the availability of helpful information and design aspects such as clarity and user-friendliness. Quantitative data from 32 active users of the app showed that 16 (50%) of patients read articles, 28 (87.5%) started journeys, and 15 (46.9%) completed exercises at any point during the study. CONCLUSIONS: The results of this mixed methods study may provide important indications for digital interventions, including ePRO assessment, for patients with cancer. From the patients' perspective, key features include intuitive design, relevant symptom items, reliable reminder functions, the translation of ePROs into clinically actionable information, clear symptom visualizations, seamless integration into electronic health records, and effective physician engagement. Both our qualitative and quantitative data show the importance of therapy-specific content. These results might increase acceptance and usage, and thereby also the clinical benefit, of future digital interventions for patients with cancer.
20 June 2026
Read appraisal →Journal of medical Internet research
Extracting and Classifying Drug Discontinuations From Estonian Electronic Health Records: Development and Validation Study
BACKGROUND: Drug adherence is crucial for chronic disease management, yet treatment discontinuation remains common due to factors such as side effects, inefficacy, or cost. These reasons are often recorded only in free-text clinical notes, making large-scale analysis difficult. While large language models (LLMs) can interpret such unstructured data more effectively than traditional natural language processing methods, few studies have systematically categorized reasons for discontinuation or identified whether the decision was initiated by the patient or the clinician, especially in low-resource languages such as Estonian. OBJECTIVE: This study aimed to assess the ability of LLMs to extract and classify reasons for drug discontinuation and identify who initiated it using Estonian electronic health records and characterize the observed discontinuation patterns and initiators for statins and antidiabetic medications. METHODS: We combined prescription data with free-text anamneses from a 10% sample of the Estonian population (2012-2019). LLMs (Llama 3.1-70B and GPT-4o) were applied to extract discontinuation phrases and reasons, classify them into a clinician-developed taxonomy, and identify who discontinued the treatment. Performance was evaluated on 100 randomly chosen cases per drug group. RESULTS: Extraction yielded 625 antidiabetic drug and 233 statin discontinuation cases. Validation confirmed a precision of 0.93 to 0.98 for extracting phrases and 0.95 to 0.96 for extracting reasons. Classification of discontinuation reasons achieved weighted F1-scores of 0.81 to 0.84, whereas classification of who initiated discontinuation achieved weighted F1-scores of 0.64 to 0.78. Adverse reactions were the most frequent reason overall, accounting for 70% (163/233) of statin discontinuations and 44.8% (280/625) of antidiabetic drug discontinuations. Regarding antidiabetic drugs, treatment inefficacy and contraindications were more common. Patients more often stopped due to adverse reactions or nonmedical reasons, whereas physicians more often initiated discontinuation for contraindications. CONCLUSIONS: LLMs can accurately extract and classify medication discontinuation reasons and show variable performance in identifying discontinuation initiators in Estonian clinical narratives. Both local and proprietary models showed promising results, enabling scalable analyses that complement structured health records. This demonstrates the potential of LLMs to unlock information from clinical notes, turning this underused electronic health record component into a valuable resource for monitoring treatment patterns and detecting adverse event signals.
19 June 2026
Read appraisal →International journal of medical informatics
Automated Flow and local LLM-Driven clinical Context Engineering: Precision colorectal cancer recurrence registry
OBJECTIVE: This study develops and validates a deployable, privacy-preserving automated tool to address the labor-intensive nature of colorectal cancer (CRC) recurrence registration. Our tool utilizes a reproducible, two-stage 'Clinical Context Engineering' workflow to mimic expert clinical reasoning and overcome the limitations of handling longitudinal clinical data ambiguity. METHODS: We retrospectively studied 3053 CRC patients (2010-2018). A local Large Language Model (LLM) (Qwen3:14B) analyzed ∼ 19,900 pathology and ∼ 43,900 imaging reports using iterative, patch-based analysis ("raw LLM"). Clinical validation rules were applied to generate "rule-based LLM" output, enhancing explainability and trustworthiness. Both automated methods and the manual Taiwan Cancer Registry (TCR) database were compared against a 20% manual reference standard (N = 602). Full prompts and validation code are provided for complete reproducibility. RESULTS: Under a strict 60-day temporal tolerance, the Rule-Based LLM achieved 90.7% accuracy, comparable to standard TCR processes (92.0%), and 77.2% sensitivity. The application of clinical validation rules significantly improved specificity from 87.7% (Raw LLM) to 93.9% (Rule-Based LLM). In time window analysis, the Rule-Based LLM identified 87.1% of recurrences within 60 days of the reference date. CONCLUSION: Our locally deployed, privacy-preserving, and explainable Clinical Context Engineering framework offers a viable, non-inferior alternative to standard TCR processes, reducing workload while maintaining data quality and fostering trust in AI-assisted cancer registry automation.
16 June 2026
Read appraisal →International journal of medical informatics
An AI-assisted framework for the ethical use of machine learning in healthcare
This study develops and evaluates ETHICS, a concise, clinician-facing ethical protocol for the routine use of machine learning (ML) in healthcare. Using ChatGPT for first-stage drafting, we generated six actionable principles - Equity and Fairness, Transparency and Patient-Centered Care, Human Oversight and Clinical Integrity, Information Privacy and Data Governance, Continuous Improvement and Sustainability, and Support and Education for Professionals - structured as a mnemonic to support uptake in time-constrained clinical settings. Outputs were treated as provisional and refined through mandatory human source verification, iterative readability optimization, multidisciplinary expert review, and scenario-based stress testing. Readability analysis showed substantial improvement from high complexity to clinician-accessible language. Expert ratings indicated strong endorsement with excellent inter-rater reliability. Implementation readiness was confirmed across five clinical scenarios, all passing predefined adequacy thresholds. The findings suggest generative AI can accelerate ethical protocol drafting, but validity and practicality depend on structured human oversight and real-world testing.
16 June 2026
Read appraisal →International journal of medical informatics
From algorithmic innovation to clinical deployment: A systematic review of methodological gaps limiting federated learning in healthcare
OBJECTIVE: Federated learning (FL) is a distributed machine learning paradigm designed to enable model training across decentralized data sources without requiring data centralization. This review critically examines FL as a methodological approach in biomedical informatics, summarizing its conceptual foundations, methodological variants, successes, and limitations, and identifying directions for future research. METHODS: We conducted a systematic methodological review following PRISMA 2020 guidelines. Searches were performed in PubMed, Web of Science, Scopus, IEEE Xplore, and Google Scholar for studies published between January 2017 and March 2025. Articles were included if they proposed, extended, or critically analyzed FL methods for biomedical or life science data. Methods were categorized by federation topology, optimization strategy, data heterogeneity handling, privacy mechanisms, evaluation design, and translational readiness. RESULTS: From 8,412 records, 97 articles met inclusion criteria. Rapid methodological innovation exists in optimization and privacy mechanisms, while support for design-time evaluation and governance remains limited. CONCLUSION: Federated learning represents a significant methodological advance in biomedical informatics but current implementations address only a subset of translational challenges. Future work must integrate study design, evaluation, interpretability, and governance into FL methods.
16 June 2026
Read appraisal →Croatian medical journal
Artificial intelligence in pathology: a framework for preserving brain capital in the diagnostic apex
AIM: To present a five-criterion calibration framework for evaluating artificial intelligence (AI) tools in pathology centered on preserving "brain capital" - the finite cognitive resources available to clinicians - and stabilizing the "diagnostic apex," the point where histological evidence meets clinical judgment. METHODS: Using cognitive load theory (CLT) and implementation science, we developed the Pathology AI Diagnostic Balance - a conceptual model mapping the physician-specimen interaction within working memory and IT infrastructure constraints - and examined it in the context of peer-reviewed case studies of implemented pathology AI tools. RESULTS: The framework identifies five criteria: Contextual Literacy (integration of clinical history), Responsibility (preservation of signing authority), Evidence Gap Analysis (closure of diagnostic uncertainty), Verification (feasibility of quality control), and Opportunity Cost (return on cognitive investment). Successful tools, such as Ki-67 quantification and breast cancer screening, reduce cognitive burden by automating routine tasks or enabling rapid verification. Conversely, tools requiring an exhaustive re-review or producing opaque recommendations deplete cognitive bandwidth, causing decision fatigue and diagnostic errors. CONCLUSION: Pathologists are positioned to exercise evaluative authority over tools that increase extraneous cognitive load. Clinical expertise represents pathologists' most distinctive contribution to AI development: defining which failure modes matter clinically, what context an AI must integrate, and whether a tool serves patient care. By prioritizing cognitive sustainability over purely technical metrics, the profession can ensure AI functions as a genuine enhancement of diagnostic capability rather than an additional burden on diagnostic workflows.
15 June 2026
Read appraisal →International journal of medical informatics
Accuracy and completeness of large language models in Epidemic keratoconjunctivitis Queries: A Comparative study
BACKGROUND: Large language models (LLMs) are increasingly applied in clinical contexts, yet their reliability in disease-specific ophthalmic domains remains insufficiently characterized. Epidemic keratoconjunctivitis (EKC), a highly contagious adenoviral ocular infection, represents a critical area where accurate diagnostic and therapeutic information is essential. This study aimed to assess and compare the accuracy and completeness of responses generated by two widely used LLMs, namely ChatGPT and Meta AI, to standardized ophthalmology questions related to EKC. METHODS: This cross-sectional expert evaluation study was conducted between January and March 2025. Thirty-four structured questions covering five EKC domains (etiology, clinical features, diagnosis, treatment, and prognosis) were developed using authoritative ophthalmology sources. Responses generated by ChatGPT (Plus subscription, web interface) and Meta AI (web interface) were independently scored by two fellowship-trained cornea specialists using a five-point Likert scale for accuracy and completeness. Inter-rater agreement was assessed using Cohen's κ and mean absolute error (MAE). RESULTS: Both models demonstrated strong overall performance, with mean accuracy and completeness scores of 4.87 and 4.90 for ChatGPT and 4.87 and 4.93 for Meta AI, respectively. Agreement between expert raters was substantial (κ range 0.78-0.88). Domain-specific analysis revealed consistently high scores across all categories, with Meta AI showing slightly higher completeness, particularly in diagnostic and symptom-related responses. CONCLUSION: LLMs exhibit consistently strong performance and domain consistency when addressing ophthalmic queries related to EKC. Their performance supports their potential use as adjunct informational and educational tools for ophthalmic conditions such as EKC, particularly for structured knowledge retrieval, treatment explanation, and patient counseling, provided that use remains under appropriate professional oversight.
15 June 2026
Read appraisal →International journal of medical informatics
Artificial intelligence applications in emergency triage for suspected acute coronary syndrome: a systematic review
BACKGROUND: Acute coronary syndromes (ACS) are time-critical conditions requiring rapid and accurate triage in the emergency department. Traditional triage may lead to delays, whereas artificial intelligence (AI) has the potential to enhance triage accuracy and efficiency. OBJECTIVE: This study aimed to synthesize and evaluate the effectiveness of AI applications in emergency triage or early clinical decision-making for suspected ACS, focusing on their impact on diagnostic accuracy, time to treatment, and operational outcomes. METHODS: This systematic review followed PRISMA guidelines. PubMed, Scopus, CINAHL, and IEEE Xplore were searched for studies published between 2020 and 2025. Eligible articles examined AI applications for triage of suspected ACS in the emergency department. Data on study design, AI models, comparators, inputs, and outcomes were extracted, and study quality was assessed using the Mixed Methods Appraisal Tool (MMAT). RESULTS: Fifteen studies from multiple countries and study designs were included. Convolutional neural networks and ensemble learning methods were the most used models. AI models generally demonstrated high diagnostic performance (AUROC 0.82-0.99), were associated with reductions in treatment times such as door-to-balloon and catheterization intervals, and showed potential to improve operational outcomes, including resource utilization and patient flow. However, findings varied across studies depending on model type, data inputs, and study design. CONCLUSIONS: AI-assisted triage for suspected ACS shows promise in supporting clinical decision-making and improving workflow efficiency. However, substantial heterogeneity and limited prospective validation suggest that findings should be interpreted cautiously. Further research is needed to confirm clinical effectiveness, generalizability, and safe implementation.
15 June 2026
Read appraisal →Journal of affective disorders
Implementing artificial intelligence (AI) to facilitate health outcomes in mood disorders: Application versus aspiration.
The burden of depressive and bipolar disorders at the individual and societal level are extraordinary and increasing. For decades, evidence-based treatments for these conditions have been established but outcomes amongst individuals with lived experience remains suboptimal. Notwithstanding calls to close gaps between current practice and best practice, there is an absence of evidence that overall health outcomes are significantly improving. Artificial Intelligence (AI) is the cornerstone of the digital revolution. Currently, AI sources (e.g., Open Source) are widely accessed by healthcare providers and persons with lived experience for search queries and decision support. The aspiration for AI-informed medical practice is to improve health outcomes by assisting in timely diagnostic detection, illness monitoring, informing treatment selection, integrating multimodality care, decreasing barriers to access and facilitating scalability to psychosocial interventions. Against this background, rigorous evidence is still needed to empirically demonstrate transformative improvement in each of the aforementioned areas. In addition, multiple ethical, technical, scientific and economic questions are not adequately answered including aspects of confidentiality and patient engagement. This short commentary endeavors to succinctly summarize the evidentiary base as it relates to the capabilities that AI offers currently, in the near and more intermediate term. The overarching aim is to provide readers with an up-to-date understanding of what aspects of AI are currently applicable versus those that are aspirational.
15 June 2026
Read appraisal →PloS one
Time-to-event ensemble machine learning approach for predicting long-term survival of abdominal aortic aneurysm patients undergoing endovascular aneurysm repair
BACKGROUND: Endovascular aneurysm repair (EVAR) for abdominal aortic aneurysm (AAA) is associated with risks such as endoleaks and late aneurysm rupture, highlighting the importance of long-term survival prediction. Despite recent advancements in machine learning (ML), predictive models utilizing time-to-event analysis remain limited for AAA patients undergoing EVAR. We aimed to develop a stacking ensemble ML model to predict long-term outcomes in EVAR-treated AAA patients. METHODS: From 2002 to 2019, a total of 12,312 patients underwent EVAR. The primary outcome was AAA-related mortality, with follow-up until December 31, 2019. Using 5 ML algorithms, we developed a model comprising 34 variables. Model performance was assessed using the time-dependent C-index and Brier score. Variable importance was evaluated through permutation-based and partial dependent plots. RESULTS: The stacking ensemble model showed the best predictive performance among the tested models (time-dependent C-index: 0.759 at 30 days, 0.716 at 365 days). The time-dependent Brier scores generally increased slightly over time but remained stable across all ML algorithms. Important predictors included age, smoking status, duration between diagnosis and surgery, household income, renal function, and blood pressure. Variable importance differed over time, and each predictor presented a nonlinear relationship with AAA-related mortality risk. CONCLUSION: The stacking ensemble ML model for time-to-event prediction identified dynamic, time-varying changes in predictor importance, providing improved risk stratification and phase-specific management after EVAR.
14 June 2026
Read appraisal →JMIR formative research
An Interactive Digital Dashboard for Patient Monitoring and Management in a Continuity of Care Centre: Development and Preliminary Usability Evaluation Study
BACKGROUND: Clinical dashboards are becoming important tools for managing and monitoring hospitalized patients across different wards. Moreover, careful attention to design, usability, and user interaction is essential for developing effective support tools for clinicians. OBJECTIVE: This study aimed to describe the development, implementation, and preliminary evaluation of an interactive dashboard for patient monitoring and management in the Continuity of Care Centre (CCC). METHODS: We developed a dashboard according to clinicians' requests and the daily workflow of case managers. First, a CCC Data Mart was created to collect all patients' information automatically extracted from the hospital's data warehouse. However, case managers had the possibility to enter additional patient information in the dashboard using a dedicated form. Moreover, CCC physicians, nurses, and administrative staff were surveyed using 2 validated questionnaires, the System Usability Scale and the Questionnaire for User Interaction Satisfaction. The Situation Awareness Index was proposed to evaluate user awareness and task efficiency. RESULTS: The first version of the CCC dashboard presented 4 panels with different types of information, both on the individual patient and on metrics related to the overall patient population. The first panel focused on patients' data, such as demographic factors, admission, transfer, discharge wards (and their dates), etc. The importance of this panel was the possibility of viewing information collected from different sources within a single interface. The other 3 panels displayed different key performance indicators for the overall patient population and presented data both in the form of tables in the second panel and graphs in the third and fourth panels. After 3 months of daily use, a total of 15 participants, 10 nurses, 2 administrative staff members, and 3 physicians, were recruited for the dashboard evaluation. The average System Usability Scale score of the dashboard was 61.5 (SD 15.7) points, which indicates "OK to good" usability, and the median score obtained with the Questionnaire for User Interaction Satisfaction was 5.77 (IQR 4-7), with the highest results in usability (mean 6.33, SD 0) and learning (mean 6.01, SD 0.39). The overall Situation Awareness Index score was 4 points, with the highest score in "familiarity of dashboard" (mean 4.73, SD 1.66 points) and "arousal support" (mean 4.6, SD 1.8 points). CONCLUSIONS: We developed an interactive dashboard for patient monitoring and management, with positive evaluations from users across different questionnaires.
14 June 2026
Read appraisal →Journal of medical Internet research
Telehealth Usability, Engagement Patterns, and Technical Infrastructure in Managing Noncommunicable Diseases Among Health Care Professionals in Brazil, Ghana, Honduras, and the United Kingdom: Multinational Cross-Sectional Study
BACKGROUND: Noncommunicable diseases (NCDs) account for over 70% of global deaths, with hypertension and diabetes serving as major contributors. The COVID-19 pandemic disrupted traditional health care services for NCDs and highlighted telehealth as a crucial alternative. Telehealth-encompassing synchronous and asynchronous electronic communication to deliver clinical services remotely-can overcome geographical barriers and enhance patient engagement. However, telehealth usability among health care professionals (HCPs) remains under-studied across low-, middle-, and high-income countries. OBJECTIVE: This study aimed to examine which telehealth engagement patterns, technical infrastructure factors, and user profiles were most strongly associated with usability among HCPs and to descriptively compare these across 4 diverse countries: Brazil (high- to middle-income country), Ghana (low- to middle-income country), Honduras (low- to middle-income country), and the United Kingdom (high-income country). METHODS: A multinational cross-sectional survey was conducted with 290 HCPs across 4 countries. Participants completed the System Usability Scale and provided data on telehealth engagement (eg, frequency, duration, and number of systems used), technical infrastructure (connection stability and support satisfaction), and their user profile (demographics, job role, and training received). Descriptive statistics summarized these patterns and usability scores. Multiple linear regression with bootstrap-based sensitivity analyses identified factors associated with telehealth usability. Given the nonprobability design, no formal inferential comparisons were made between countries. Instead, observed patterns were reported descriptively. RESULTS: Higher telehealth usability scores were associated with greater connection stability (b=5.06, 95% CI 3.06-7.05), higher satisfaction with online support information (b=5.02, 95% CI 3.27-6.75), more frequent use (b=3.05, 95% CI 1.36-4.73), longer duration of use (b=1.59, 95% CI 0.49-2.68), and being a physician by profession (b=3.82, 95% CI 0.23-7.40). Average usability scores were highest among users in Ghana (mean 79.75, SD 14.19) and the United Kingdom (mean 79.00, SD 14.71), followed by Brazil (mean 72.01, SD 14.62) and Honduras (mean 63.09, SD 15.57). According to System Usability Scale guidelines, scores corresponded to "good" usability for users in Ghana, the United Kingdom, and Brazil and were below the "good" threshold for users in Honduras. While most users in Ghana (97/111, 87.4%), Honduras (31/38, 81.6%), and Brazil (57/80, 70.4%) reported using only 1 telehealth system, two-thirds of UK users (40/60, 66.7%) reported using 2 or more systems. User profiles also varied; prepandemic use was highest in Ghana (84/111, 75.7%) and lowest in Honduras (7/38, 18.4%). Other engagement patterns across countries were reported. CONCLUSIONS: Telehealth usability is driven by technical infrastructure reliability, a robust online support infrastructure, and an "experience effect" from frequent and long-term engagement. Descriptive differences in engagement patterns and infrastructure highlight the need for tailored strategies to address setting-specific challenges. These are essential to optimize telehealth integration and improve health care outcomes for patients with NCDs worldwide.
14 June 2026
Read appraisal →Journal of the American College of Surgeons
Improving Trauma Triage Accuracy with Large Language Models: A Comparison to Human Expert Decisions
BACKGROUND: Accurate prehospital trauma triage and communication determine morbidity, mortality, and system efficiency. Advancements in large language models (LLMs) offer an opportunity to improve triage and yet to be implemented in prehospital trauma triage. STUDY DESIGN: This retrospective cohort study evaluates LLM performance in trauma triage and accuracy of prehospital tele-communication. Of 410 pediatric activations at a Level I center (January 2023 to May 2025, IRB no. 00009569), 133 activations with emergency medical service recordings, human-generated trauma pages, and injury severity scores were analyzed. Audio was transcribed with OpenAI Whisper. Structured "Essential Transcripts" were generated with named entity recognition. Entity ablation tested redaction of triage parameters on accuracy. In a prospective arm, trauma surgeons reviewed emergency medical service transcripts and triaged activations pre- and post-LLM exposure. Cribari criteria defined over and undertriage. McNemar's test and 95% CIs assessed paired differences in accuracy. RESULTS: The primary endpoint: LLM undertriage demonstrated modest improvement; 4.8% (3 to 8.2) vs 5.1% (3.1 to 9.3, p = 0.73, Bonferroni p = 1). For secondary endpoints, LLM triage outperformed human clinicians: 83.5% accuracy (80.6 to 90.6) vs 78.9% (73.9 to 82.6, p < 0.01, Bonferroni p < 0.01), with overtriage 58.6% (51.4 to 73.7) vs 71.8% (p < 0.05, Bonferroni p < 0.09). "Essential Transcripts" reduced transcript length by 80.8% (p < 0.001) while preserving accuracy (81.9%, 76.6 to 87.5, p < 0.001, Bonferroni p < 0.05). Entity ablation had marginal effect on triage. In prospective evaluation, human triage accuracy improved after LLM exposure (73.7% [69.8 to 77.2]) to 75.8% ([71.9 to 79.2], p = 0.04, Bonferroni p = 0.12), significantly improving the odds of a correct triage decision (odds ratio 2.57, 95% CI 1.39 to 6.83, p < 0.01). CONCLUSIONS: LLMs achieve triage accuracy comparable to trauma staff in retrospective review of pediatric trauma. Further validation is needed to assess clinical outcomes, generalizability, and user acceptance before widespread deployment.
14 June 2026
Read appraisal →Systematic reviews
Using full agreement across multiple large language models for title-and-abstract screening in systematic reviews: a proof-of-concept
BACKGROUND: The exponential growth of scientific literature poses significant challenges for conducting systematic reviews, particularly in the labor-intensive title-and-abstract screening phase. This study examines the feasibility of using multiple large language models (LLMs) for title-and-abstract screening in systematic reviews. METHODS: We propose and evaluate a full-agreement approach using three commercially available LLMs from different model families (ChatGPT, Gemini, and Claude), in which automated classification decisions are only accepted when all three models assign the same label. This approach was examined across six datasets to assess its effectiveness. A structured workflow was developed to support implementation without requiring specialized technical expertise. Additionally, a stop criterion was introduced to ensure that LLM-based classification is only applied when predefined performance thresholds are met. RESULTS: Across the six datasets, approximately four in five abstracts received full agreement across models and could therefore be classified automatically. For this subset of abstracts, classification performance was consistently higher than that of previous automated approaches using LLMs, with statistically significant improvements in all prespecified performance metrics. CONCLUSIONS: A full agreement approach across three LLMs may offer a promising and conservative strategy for title-and-abstract screening in systematic reviews by automating concordant decisions while reserving human review for discordant cases. As a proof-of-concept, the present findings support this approach as a possible workflow for reducing screening burden in the face of continued growth in the scientific literature, although its broader generalizability remains to be established.
14 June 2026
Read appraisal →Journal of medical Internet research
Experiences of Home-Dwelling Older Adults and Their Family Caregivers With Digital Health Services: Qualitative Study
BACKGROUND: As the population ages, demand for health services among older adults and caregivers increases. Digital health services help meet this demand. However, acceptance and experiences differ between older adults and caregivers. OBJECTIVE: This study aims to explore the specific experiences of home-dwelling older adults and their caregivers with digital health services, identify their willingness to use these services and the factors influencing their adoption, and identify their service needs to inform the subsequent design and optimization of digital health products and services. METHODS: From December 2023 to February 2024, a descriptive qualitative study was conducted. Researchers used purposive and maximum variation sampling to recruit 18 home-dwelling older adults and 17 family caregivers, including 10 matched dyads. Participants came from 7 community health service centers and a tertiary hospital in Hefei, China. Open-ended, semistructured, face-to-face interviews were conducted separately. Data were analyzed using conventional content analysis. RESULTS: Four main themes emerged. The first were the application characteristics, including usage situations and preferred functions. The second theme described specific service experiences, both positive and negative. The third was the influencing factors. Promoting factors included health needs, user experience, and subjective norms. Obstructive factors differed between groups and included low digital literacy, weak economic foundation, service security concerns, and lack of time. The fourth theme covered suggestions and expectations for improvement. Older adults appreciated the advantages but often had suboptimal experiences due to complex designs and threats to personal dignity. Conversely, caregivers valued efficiency but reported low use, hindered by time constraints and concerns about privacy and safety. CONCLUSIONS: This study reveals a critical gap between the potential and actual use of digital health services. Older adults face barriers with usability and digital literacy. Caregivers struggle to integrate these services and have trust issues. Future optimization requires action at several levels. Government support is needed through policies, authoritative platforms, and subsidies. Service providers should offer age-friendly, personalized design, and continuous support. Social support systems are also essential, such as digital reciprocity from family and peer help. These strategies are critical for enhancing user experience, bridging the digital divide, and enabling both older adults and their caregivers to fully benefit from digital health technologies.
14 June 2026
Read appraisal →JMIR nursing
Assessing Digital Literacy Among Nursing Faculty Under the Artificial Intelligence-Enhanced Technological Pedagogical Content Knowledge Framework: Cross-Sectional Survey Analysis
BACKGROUND: Digital literacy has become a core competence for nursing faculty to adapt to the digital transformation of education, improve teaching quality, and cultivate innovative nursing talents. OBJECTIVE: This study aimed to investigate the digital literacy status and influencing factors among nursing faculty in colleges and universities in Fujian province, China. METHODS: A cross-sectional survey analysis was conducted among 339 participants from July 2023 to June 2024 by selecting nursing faculty from 3 medical universities and 5 colleges in Fujian province. A general information questionnaire and a teacher digital literacy questionnaire (designed based on the artificial intelligence-enhanced technological pedagogical content knowledge theoretical model) were used to conduct the survey. The questionnaires included the following dimensions: an intelligent integration ethics layer (digital social responsibility), attention to the moral boundaries of applying technology; an awareness layer (digital awareness), the willingness to actively adapt to technological changes; a knowledge layer (digital technology knowledge and skills), mastery of the integration point of intelligent tools and subject teaching; an ability layer (digital application), the ability to design intelligent teaching plans; and a thinking layer (professional development; innovative thinking for the critical integration of technology). Statistical analysis included descriptive statistics, a 1-way ANOVA, Pearson correlation analysis, and multiple linear regression (using SPSS 28.0 software). RESULTS: The overall digital literacy of the nursing faculty was at a medium level (mean score 101.92, SD 16.47), with the highest score in digital awareness (mean 20.21, SD 9.43) and the lowest score in digital technology knowledge and skills (mean 9.68, SD 2.92). For the dimension of digital technology knowledge and skills, both age (β=-.142; P=.009) and years of teaching (β=-.147; P=.006) were significant negative predictors. Regarding digital application, age was found to be a significant negative predictor (β=-0.124; P=.02) but teaching experience was a positive predictor (β=0.123, P=.02). Similarly, age also had a significant negative impact on the professional development dimension (β=-.153; P=.005) and the overall digital literacy level (β=-.136; P=.01). CONCLUSIONS: The digital literacy of nursing faculty is at a medium level, with especially low scores in technical knowledge. Age and teaching experience are key influencing factors. Recommendations to promote balanced development of awareness and skills include providing targeted training for senior faculty, integrating courses with clinical practice, optimizing the allocation of digital resources, promoting interdisciplinary cooperation, and strengthening ethical and safety training.
14 June 2026
Read appraisal →BMJ open
Barriers and facilitators to digital health tool adoption for hypertension management: systematic review of qualitative studies
OBJECTIVES: Digital health interventions (DHIs) show considerable promise in supporting hypertension self-management by promoting preventative care and self-monitoring. While their efficacy is increasingly evident, the long-term uptake, acceptance and sustained engagement with these tools are frequently challenged by issues such as usability, trust and varying user experiences. This review aims to synthesise qualitative evidence to identify barriers and facilitators and the key factors that impact the adoption, acceptance and engagement with DHIs for hypertension self-management. DESIGN: Systematic review of qualitative literature using thematic analysis following Cochrane's qualitative and implementation methods guidance. DATA SOURCES: PubMed, PsycInfo, Web of Science and the Cochrane Library were searched in February 2025. ELIGIBILITY CRITERIA FOR SELECTING STUDIES: The searches included relevant qualitative and mixed-methods studies on the use of digital devices for hypertension management, which described the barriers and facilitators associated with these tools. We included studies published from 2015 to 2025 to capture relevant evidence. Only studies published in English with a qualitative approach were included. RESULTS: From an initial 10 943 identified publications, 15 met our inclusion criteria, primarily originating from Europe and the USA, exploring diverse racial and ethnic group experiences. Our thematic synthesis revealed 7 analytical and 22 descriptive themes detailing barriers and facilitators encountered by patients with hypertension, healthcare providers (HCPs) and caregivers. These themes covered technology utilisation, design components, linguistic and cultural relevance, healthcare factors, trust and credibility and interpersonal interactions. CONCLUSION: Our analysis underscores that factors such as the usability, design and relevance of social support profoundly influence the uptake and acceptance of DHIs in hypertension self-management among patients, caregivers and HCPs. PROSPERO REGISTRATION NUMBER: CRD42023480389.
14 June 2026
Read appraisal →Family practice
Current state of electronic problems lists in primary care: a rapid scoping review
BACKGROUND: Electronic problem lists (PLs) are central to the problem-oriented medical record and increasingly underpin clinical decision support (CDS), interoperability, and emerging artificial intelligence applications in primary care. However, persistent concerns regarding PL accuracy, completeness, and governance limit their clinical value. OBJECTIVE: To update the foundational framework proposed by Hodge and Narus by synthesizing contemporary evidence on electronic PLs in primary care and primary care-relevant settings. METHODS: We conducted a rapid scoping review following Joanna Briggs Institute guidance and PRISMA-ScR reporting standards. MEDLINE (Ovid) was searched for peer-reviewed studies published between March 2016 and May 2026. Eligible studies examined digital PLs or extractable PL-related practices relevant to primary care. Findings were synthesized using a hybrid inductive-deductive approach and mapped to the seven Hodge and Narus themes. RESULTS: A total of 103 studies were included. Across settings, PLs were widely recognized as foundational for longitudinal care, safety, and coordination. However, incomplete, outdated, or inconsistently maintained PLs remained common. Persistent challenges included unclear ownership, workflow misalignment, and variability in what constitutes a "problem." Emerging technologies-including CDS, natural language processing, and machine learning-were increasingly used to support PL generation, curation, reconciliation, and organization, with greatest benefit when embedded within clinician-led workflows. CONCLUSION: Despite technological advances, the clinical value of electronic PLs in primary care continues to depend on governance, workflow integration, and shared accountability. Updating the Hodge and Narus framework highlights the need for sociotechnical approaches to PL improvement as PLs become integral to CDS and artificial intelligence-enabled care. Problem lists (PL) are a key part of electronic health records. They provide clinicians with a quick overview of a patient's important health issues—such as long-term conditions, past diagnoses, and current concerns—and help support safe and informed care. When PLs are accurate and up to date, they can improve communication and decision-making. However, in everyday clinical practice, many PLs are incomplete, outdated, or difficult to maintain. As electronic health records are increasingly used to support clinical decision support tools and new digital technologies, the quality of PLs has become more important than ever. Poor-quality PLs can lead to missed information, duplicated work, or unsafe care decisions. To better understand how PLs are working in modern healthcare, we conducted a rapid scoping review of research published between 2016 and 2026. We examined studies that looked at how PLs are used, how accurate they are, what challenges clinicians face in keeping them updated, and what solutions have been proposed. In total, 103 studies were included. Across these studies, several common problems were identified. Important health conditions were often missing or out of date, and clinicians were frequently unsure who was responsible for updating the PL. Time pressures and busy clinical workflows made regular maintenance difficult. Some studies explored new digital tools, such as automated text analysis, to help identify or update problems, but these approaches are still developing and require careful evaluation. Improving PLs will require clearer standards, shared responsibility among healthcare teams, and EHRs that make updating information easier. Thoughtful use of new technologies may also help reduce clinician workload while improving patient safety. This review highlights where PLs are working well and where improvements are still needed, offering guidance for future research, technology design, and improvements in primary care practice.
14 June 2026
Read appraisal →Journal of neurology, neurosurgery, and psychiatry
Impact of an artificial intelligence-driven triage system on workflow and transfer efficiency: stratified analysis of 4548 admissions to four thrombectomy hubs receiving transfers from sixty spokes.
BACKGROUND: We aimed to evaluate the impact of implementing an artificial intelligence (AI)-enabled acute ischaemic stroke triage system on workflow efficiency and transfer optimisation in a large academic healthcare network. METHODS: A prospectively maintained database was reviewed comparing equivalent time periods before and after AI-enabled triage platform implementation (January 2021-December 2022). The primary analysis compared workflow metrics between AI-enabled and non-AI spokes during the same calendar period (2022) to control for temporal confounding. Benjamini-Hochberg correction was applied for multiple comparisons, and analyses were adjusted for age and baseline National Institutes of Health Stroke Scale. Evaluated outcomes included door-in-door-out (DIDO) times, door-to-puncture (DTP) times, endovascular therapy (EVT) utilisation rates, cost analysis and clinical outcomes at discharge. RESULTS: The study included 4548 admissions with 844 EVT patients (394 pre-implementation, 450 post-implementation) across four hub centres. In the primary same-period analysis (2022), AI-enabled spokes demonstrated significantly shorter DIDO times compared with non-AI spokes (median 103 (92-118) vs 134 (103-162) min; adjusted difference -41.6 min (95% CI -60.9 to -24.1); p<0.001, Q<0.001) and shorter DTP times (21 (14-43) vs 40 (18-65) min; adjusted difference -10.9 min (95% CI -17.9 to -3.7); p=0.003, Q=0.009). A difference-in-differences analysis demonstrated that DIDO improvements were specific to AI-enabled spokes (-27 min; 95% CI -62 to -4; p=0.029). EVT utilisation was also significantly higher in AI-enabled versus non-AI spokes where AI-enabled spokes had increased EVT rates by +17.8% (39.3% to 57.1%) compared with +1.1% in non-AI spokes (41.3% to 42.4%, Pinteraction=0.006). DTP improvements were more pronounced at community hubs (86 (48-108) to 51 (22-77) min; adjusted difference -24.9 min; p=0.021, Q=0.041) compared with academic hubs (60 (23-87) to 55 (22-73) min; adjusted difference -15.5 min; p<0.001, Q=0.002). Subgroup analyses demonstrated consistent DIDO benefits across age, stroke severity and sex strata with no significant treatment effect heterogeneity (all P-interaction >0.05). Probabilistic cost analysis estimated savings of $3.6 million (95% CI $1.5M to $6.1M) per 1000 AI-enabled spoke transfers. Clinical outcomes, including functional status and mortality at discharge, were similar between groups (all Q>0.05). CONCLUSION: Implementation of an AI-enabled triage platform was associated with significant reductions in workflow times and increased EVT utilisation, with effects specific to AI-enabled spokes rather than secular trends alone. The proportion of transfers who did not proceed to EVT decreased in AI-enabled spokes, though counterfactual outcomes for non-transferred patients remain unknown. Clinical outcomes at discharge were unchanged.
14 June 2026
Read appraisal →Journal of medical Internet research
Uptake of Clinical Decision Support Systems Among Health Care Professionals in Six European Countries and the United States: Cross-Sectional Survey Within the I-CARE4OLD Project
BACKGROUND: The use of Clinical Decision Support Systems (CDSS), such as clinical decision rules, algorithms, or machine learning-based applications, has gained attention in recent years. However, their adoption and effectiveness may vary across different health care systems and settings. For a CDSS to be adopted, it must effectively address the practical issues encountered by professionals; however, little research has been done to identify these needs and requirements. OBJECTIVE: This study aims to describe and compare the current use of various decision-support and prediction tools in long-term care for older people across health professionals from 6 European countries and the United States. METHODS: This study analyzed survey data from a CDSS pilot study in a purposive sample of health professionals working with older adults with complex chronic conditions from six European countries and the United States. The survey included participants' general background information, their current use of decision support tools, and their attitudes on the potential benefits of CDSS. About 20 participants were sampled per country. Closed responses were analyzed using correlation coefficients and regression models, while open-ended responses were clustered in a qualitative manner, categorizing each response. RESULTS: A total of 151 professionals (mean age 45.5, SD 11.6 years, 71.5%, 108/151 female) participated in the pilot study. Most participants were physicians (85/151, 56.3%) or nurses (57/151, 37.7%). About 51% (78/151) of the participants reported using CDSS, while 22.4% (34/151) used predictive CDSS, showing important variation across samples from the seven countries. The regression model for comfort with technology showed a positive association for openness to new technologies (β=0.622; P<.001), although an inverse significant association was found for age (β=-0.022; P<.001). No significant associations were found for the actual use of CDSS. Participants reported using CDSS mainly for diagnostic purposes or for guideline implementation, not aimed at prognostic information. In contrast, examples of prognostic tools were most frequently mentioned by respondents as being valuable improvements to clinical practice. CONCLUSIONS: While some countries' samples reported well-integrated digital health infrastructures and higher CDSS adoption rates, others still face challenges in implementing these. However, we found multiple examples of emerging tools, and at the same time, an important demand for predictive CDSS. Our findings highlight the need for improvement of current CDSS implementation and both development and implementation of particularly predictive CDSS.
14 June 2026
Read appraisal →Journal of medical Internet research
Telehealth Scale and Artificial Intelligence Adoption Tiers Across Clinical and Operational Domains in US Hospitals: Cross-Sectional Study
BACKGROUND: Telehealth expansion and artificial intelligence (AI) adoption are often described as parallel dimensions of health system digital transformation. However, whether telehealth scale is associated with hospital AI adoption and whether this relationship varies across hospital settings remain unclear. OBJECTIVE: This study examined the association of telehealth scale with clinical and operational AI adoption tiers in US hospitals and assessed whether these patterns differed by telehealth reporting behavior and geography. METHODS: This cross-sectional study included 6173 US acute care hospitals using linked 2024 American Hospital Association Annual Survey and Information Technology Supplement data and 2023 Healthcare Cost Report Information System data. Telehealth scale was parameterized using log-transformed telehealth volume, a telehealth nonreporting indicator, and a reported-zero telehealth indicator. Clinical and operational AI adoption tiers were derived from hospital-reported AI capability items and classified into 3 tiers. Both outcomes were modeled using multioutput gradient-boosted tree classifiers, and model behavior was interpreted using Shapley additive explanations, partial dependence plots, and stratified analyses by the Core-Based Statistical Area category. RESULTS: Telehealth volume was the strongest predictor of both clinical and operational AI adoption tiers and had a larger contribution to the clinical AI model. Telehealth nonreporting was common, occurring in 57% (3521/6173) of hospitals, and was concentrated among hospitals in the lowest clinical AI adoption tier, accounting for 91.4% (3145/3441) of hospitals with no reported clinical AI adoption. Higher telehealth volume was associated with a steep increase in predicted clinical AI adoption tiers at lower telehealth volumes, followed by a plateau at higher volumes. At similar telehealth volumes, rural hospitals showed weaker telehealth-attributed contributions to predicted clinical AI adoption tiers than metropolitan hospitals. Supplementary analyses suggested that telehealth reporting status and telehealth intensity reflected related but distinct structural processes. CONCLUSIONS: Telehealth scale was strongly associated with hospital AI adoption tiers, especially clinical AI adoption tiers. These findings suggest that telehealth capacity may serve as a practical hospital-level marker of broader digital readiness for AI adoption, but the cross-sectional design does not establish whether telehealth expansion precedes or causes AI adoption. Hospitals with telehealth nonreporting and rural hospitals may face additional structural barriers that limit the translation of digital capacity into AI maturity. Policies to reduce inequities in hospital AI adoption may therefore need to pair telehealth expansion with implementation support, interoperability capacity, and organizational resources.
14 June 2026
Read appraisal →JMIR public health and surveillance
Digital Health Center App for Community and National Malaria Surveillance in Cambodia: Implementation Case Study
BACKGROUND: By 2015, the emergence and dissemination of multidrug-resistant Plasmodium falciparum in the Greater Mekong Subregion threatened regional and global malaria control efforts. In response, Greater Mekong Subregion countries committed to malaria elimination by 2030, with strengthened surveillance as a strategic pillar. In 2017, Cambodia introduced an elimination-oriented digital Malaria Information System (MIS). Its health center app enables real-time, geo-located, case-based malaria reporting across primary health centers, and is fully integrated with the MIS. OBJECTIVE: This study aimed to evaluate the real-world national implementation of Cambodia's Android-based health center app, considering coverage, fidelity, timeliness, and data use, and their effects on malaria surveillance performance, case management, programmatic response, and public health outcomes. METHODS: System performance and public health use were assessed using system-generated metadata, national surveillance data, and user surveys. Operational indicators included technical performance, data completeness, and reporting timeliness, alongside surveillance outcomes such as case notification, classification, reactive case detection, and foci investigation. Nationwide user experience was measured via a survey of 761 health centers across 21 provinces, with in-depth structured surveys at 9 health centers in 3 provinces. Descriptive analyses evaluated system functionality, contribution to malaria surveillance and response, and usability among frontline health workers. RESULTS: The health center app demonstrated strong technical performance, with rapid loading and resilient data transmission under low-bandwidth conditions, supporting reliable reporting in resource-constrained settings. Integrated real-time dashboards provided analytics for case management, surveillance monitoring, risk stratification, and targeted public health interventions. Data completeness remained high (99%, 89/90 fields in 2024), demonstrating consistent routine use even as case incidence declined. Between January 1, 2025, and July 31, 2025, 69 malaria cases were reported nationally (23 locally acquired, 7 domestically imported, and 39 internationally imported). Of these, 95.7% (66/69) were notified and classified within 1 day. Reactive case detection was completed within 3 days for all 21 eligible cases, and 16 of 19 eligible foci received a response within 7 days, indicating strong operational responsiveness. User surveys showed 96.3% (733/761) of health centers were satisfied or very satisfied, 90.1% (686/761) reported rare or no technical issues, and 91.7% (698/761) found the app easy to navigate. Operational challenges included limited internet connectivity, transport to remote areas, and electricity interruptions. In-depth surveys confirmed high uptake, confidence in reporting, and routine use of surveillance data, although gaps in local analytical capacity were identified. CONCLUSIONS: Developed and managed locally to enhance sustainability, the MIS drove significant reductions in malaria case incidence, with the health center app contributing timely, complete, structured reporting at the point of care. Public health responses were facilitated by real-time analysis, targeted interventions, and decentralized decision-making. User engagement was sustained as malaria cases declined, and further enhancements are planned to ensure seamless transition to postelimination surveillance, reducing the risk of malaria reestablishment in Cambodia.
13 June 2026
Read appraisal →Critical care (London, England)
Large language models for optimizing clinical trial recruitment in ICUs: application to ventilator-induced diaphragm dysfunction
BACKGROUND: Ventilator-induced diaphragm dysfunction (VIDD) is a frequent and under-recognized consequence of prolonged mechanical ventilation in intensive-care unit (ICU) patients. Identifying eligible candidates for clinical trials targeting VIDD remains a major operational challenge. This study evaluates the use of large language models (LLMs) to automate patient prescreening from ICU discharge summaries and estimate recruitment capacity for a future phase 2 trial. METHODS: We developed an LLM-based prescreening pipeline to assess trial eligibility criteria from ICU discharge summaries, which was deployed to screen all 2024 ICU stays. Stays that were flagged as potentially eligible underwent expert adjudication. An enriched set of 50 ICU stays was independently annotated by six clinicians to define a reference standard, which was used to evaluate criterion-level model performances using F1-scores. RESULTS: The best-performing model was GPT-OSS:120B with a criterion-level F1-score of 0.82. When applied to consecutive 1,342 ICU stays from Montpellier University Hospital in 2024, the selected model identified 532 patients with ≥ 3 days of mechanical ventilation. After applying exclusion criteria, 185 patients remained potentially eligible. Expert review confirmed 133 patients as eligible, resulting in a positive predictive value of 72% (95% CI 65-78). The LLM-assisted workflow resulted in an estimated 86% reduction in clinician review time. The LLM-based prescreening pipelines achieved criterion-level F1 scores ranging from 0.73 to 0.82, with GPT-OSS:120B demonstrating the highest performance. CONCLUSIONS: LLM-based prescreening offers a promising approach for identifying trial candidates in critical care, prioritizing candidates for clinician review. Future deployments should include targeted expert validation and ongoing monitoring to ensure safety and generalizability.
13 June 2026
Read appraisal →JMIR nursing
Association Between Electronic Health Record-Based Nursing Workload and Turnover: Retrospective Cohort Study
BACKGROUND: Nurse turnover remains a major challenge for health systems, yet objective, scalable measures of workload that predict turnover are limited. Electronic health record (EHR) audit logs offer a potential data source to quantify nursing work patterns. OBJECTIVE: This study aimed to evaluate the association between EHR-derived measures of nursing workload and turnover among inpatient nurses. METHODS: We analyzed work-related activities of staff nurses from medical and surgical inpatient units at a large academic medical center from January 1, 2022, to December 31, 2022. Data included deidentified demographics (age, sex, years since licensure, and service group), shift characteristics (number of shifts worked, proportion of night shifts, timing, and location), and measures of work activities derived from EHR audit logs. Audit logs were used to develop measures related to nurse workload, including (1) nurse activities (information review, medication administration, alert management, navigation, documentation, and communication), (2) patient load (based on the number of unique patient charts accessed), and (3) cognitive load (based on the number of patient switches). For nurses who left, we excluded the 6 weeks immediately preceding termination (washout period) and measured workload during the preceding 6-week period; for those who remained, a random 6-week working period was selected. Associations between workload measures and turnover were assessed using mixed-effects logistic regression, adjusting for demographics and shift-related characteristics. RESULTS: Among 432 nurses (n=363, 84% female; median age 27, IQR 23-36 years), contributing 6812 shifts and approximately 13 million audit log actions, 84 (19%) left the institution in 2022. A higher proportion of medication administration actions was associated with higher odds of turnover (odds ratio [OR] 2.20, 95% CI 1.36-3.54), whereas greater alert engagement (OR 0.48, 95% CI 0.32-0.72) and more years since licensure (OR 0.57, 95% CI 0.38-0.84) were associated with lower odds of turnover. CONCLUSIONS: EHR-derived workload measures, particularly greater medication administration burden and lower alert engagement, were independently associated with the risk of nurse turnover. Patterns of EHR use may help identify nurses at higher risk of leaving an institution and can potentially inform targeted workforce retention strategies.
13 June 2026
Read appraisal →Med (New York, N.Y.)
Rethinking healthcare data interoperability in the age of large language models
Electronic health records contain extensive real-world clinical data, but their effective use is hindered by data heterogeneity and interoperability challenges. Traditional post hoc standardization is costly and labor-intensive and reduces data granularity. Large language models enable analysis of unstructured data without full harmonization but lack precision for some tasks. We propose a hybrid strategy that combines large-language-model-based analysis of legacy data with prospectively standardized data, offering a scalable alternative that challenges the need for retrospective data harmonization and improves interoperability.
13 June 2026
Read appraisal →Journal of the American College of Surgeons
Governance Framework for Safe and Ethical Implementation of Artificial Intelligence in Surgery: A Modified Delphi Consensus
BACKGROUND: Artificial intelligence (AI)-enabled clinical decision support systems (CDSS) demonstrate performance comparable or superior to human experts in certain tasks. However, their integration into surgical practice faces a significant implementation gap, alongside ethical, privacy, and legal concerns. Clear governance frameworks are needed to guide their responsible adoption in surgery, to prevent inconsistent application, care quality variation, and exacerbation of algorithmic bias. Here, we establish a systematic, evidence-based, and consensus-driven framework to guide the ethical, effective, and sustainable adoption of AI-enabled CDSS in surgery. STUDY DESIGN: A systematic literature review was conducted of PubMed, Cochrane Library, Medline, and Embase databases until 2024 to identify key governance themes. The themes informed the generation of candidate items, which were then refined through a multiround expert panel consensus process using a modified Delphi approach to produce the final framework. RESULTS: Thematic analysis of 80 full-text articles meeting inclusion criteria identified 4 overarching themes for AI governance: (1) technical prerequisites and model design; (2) clinical implementation and human factors; (3) ethics, safety, and trustworthiness; and (4) bias, fairness, and equity. Panel consensus evaluation resulted in the development of a 19-item framework. CONCLUSIONS: The consensus-driven framework presented here provides foundational guidance essential for navigating the complexities of implementing AI-enabled CDSS safely and ethically in surgery. Addressing the considerations outlined across these 4 core themes can facilitate the responsible adoption of AI, accelerating the transition toward an advanced, data-driven surgical practice while mitigating potential risks.
13 June 2026
Read appraisal →JMIR medical informatics
Machine Learning for Intraoperative Bleeding Prediction in Patients Undergoing Surgery: Scoping Review
BACKGROUND: Intraoperative bleeding is a critical event that impacts surgical safety and patient outcomes. Machine learning (ML) has demonstrated potential in prediction tasks, yet its methodological rigor and clinical translation face challenges. OBJECTIVE: This scoping review aims to systematically synthesize the current state of development, performance, and validation of ML models for predicting intraoperative bleeding, and to identify key barriers to their clinical implementation. METHODS: Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines, we systematically searched 7 databases (PubMed, Web of Science, Embase, CINAHL, CNKI [China National Knowledge Infrastructure], Wanfang, and VIP [China Science and Technology Journal Database]) from their inception to April 2025. Moreover, 2 reviewers (SY and PZ) independently screened studies, extracted data using the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies (CHARMS), and assessed the risk of bias using the Prediction Model Risk Of Bias Assessment Tool (PROBAST). A narrative synthesis was used for data analysis. RESULTS: Out of 2651 screened records, 23 studies were included (sample sizes ranging from 48 to 48,543). Tree-based ensemble models (eg, random forests and extreme gradient boosting) were the most frequently used (16/23, 70%), followed by logistic regression (13/23, 57%), and deep learning (11/23, 48%). Model discrimination varied widely (mean area under the curve [AUC] 0.82, SD 0.08, range 0.63-0.93). Integration of multimodal data (electronic health records+imaging) was associated with higher performance. However, model validation was often inadequate; only 6 studies (6/23, 26%) performed external validation, and performance often declined (eg, AUC decreased from 0.85 to 0.63 in 1 study). Reporting exhibited selective bias; AUC was commonly reported (19/23, 83%), whereas key classification metrics, such as calibration (10/23, 43%) and precision (4/23, 17%), were often omitted. PROBAST assessment indicated a high risk of bias in all included studies (23/23, 100%). CONCLUSIONS: While ML models demonstrate technical promise for predicting intraoperative bleeding, our PROBAST assessment revealed a universally high risk of bias across all included studies. This fundamental methodological limitation, coupled with a severe lack of external validation and poor transparency in reporting, severely constrains the current clinical reliability of these models. Future research must prioritize prospective multicenter validation, adherence to Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) reporting guidelines, and enhanced model interpretability to bridge the gap toward clinical utility.
13 June 2026
Read appraisal →PloS one
Mumtathil: Automatic PDPL compliance identification system of Arabic privacy policies documents.
In the digital era, the extensive collection and use of personal data by websites and applications has prompted various countries to implement data protection laws. The Kingdom of Saudi Arabia introduced the Personal Data Protection Law (PDPL) to regulate the handling of personal data. However, manually assessing compliance with these laws is time-consuming, and costly. This paper presents an automated system named Mumtathil (مُمتثِل) that evaluates the compliance of Arabic privacy policies with the PDPL using multi-label classification (MLC) techniques. Our experiments demonstrated that the Support Vector Machine (SVM) outperformed other machine learning models and transformer-based models like CamelBERT and AraBERT, achieving an F-score of 92%. The system addresses dataset imbalance by augmenting the Saudi Privacy Policy Dataset with recent privacy policies and synthetic data generated using ChatGPT. The Mumtathil tool simplifies compliance assessment, reducing time, cost, and effort while enhancing transparency and data protection in the digital landscape.
12 June 2026
Read appraisal →JMIR formative research
Usability, Relevance, and User Engagement of a Government-Sponsored and Community Co-Designed Digital Mental Health Website During the COVID-19 Pandemic: A Pilot Study on Latino/a/x Adults
BACKGROUND: Latino/a/x adults have higher rates of unmet mental health needs than other racial and ethnic groups. One promising solution to help bridge this gap in care is digital mental health tools. Digital tools, such as self-help websites, have demonstrated the ability to enhance mental health literacy, reduce stigma, and improve mental health symptoms. Despite the potential benefits, engagement remains a critical challenge, and there has been a large oversight of unique considerations for Latino/a/x adults as end users. OBJECTIVE: Guided by the Technology Acceptance Model and the Behavioral Model for Vulnerable Populations, the study's overarching objective is to characterize within-group variation of Latino/a/x adults' engagement with a government-funded, prevention-focused mental health website that was co-designed with community partners during the COVID-19 pandemic. METHODS: The Together for Wellness/Juntos Por Nuestro Bienestar (T4W/Juntos) website offered free digital mental health resources to help Californians cope with the COVID-19 pandemic. A pilot evaluation of the website involved baseline and 4-week follow-up surveys about demographics, behavioral health needs, and overall website user experience. This current subanalysis focused on a stratified sample of Latino/a/x adult participants (baseline N=131; baseline and follow-up n=68). The baseline sample was mostly female (106/130, 81.5%); 66.9% (87/130) preferred to use the website in English and 30% (39/130) preferred Spanish. Behavioral health needs were assessed using the Patient Health Questionnaire-2, Generalized Anxiety Disorder 2-item scale, and a COVID-19 stressors checklist. We measured usability, comfort using the website, relevance of the website, and past-month use of resources. Data were analyzed using ordered and standard logistic regression methods. RESULTS: Latino/a/x adults who preferred using the website in English (odds ratio [OR] 13.76, P<.001) compared with Spanish were more comfortable using the website. Compared with adults aged 18-30 years, adults aged 41-50 years had significantly lower odds of agreeing that the website was easy to use. Sensitivity analyses revealed that participants who found the website easier to use (OR 2.22, P=.001) and those with greater behavioral health needs (OR 1.22, P=.045) were more likely to perceive the website's topics as relevant. Participants with higher behavioral health needs at baseline were more likely to use the website and engage with resources for anxiety or stress at follow-up (OR 1.42, P=.047). CONCLUSIONS: This study addresses gaps in understanding Latino/a/x adults' experiences with a prevention-focused mental health website. The language-based disparity in comfort highlights the need to significantly improve the user experience for Latino/a/x Spanish speakers. Still, the website can be a helpful resource for Latino/a/x adults with high behavioral health needs, bridging a critical gap in support. A collaborative approach to developing resource-rich websites with trusted community organizations is vital for effectively reaching Latino/a/x communities and tailoring resources to address their unique needs.
12 June 2026
Read appraisal →BMJ open
Prediction of ICU length of stay, hospital discharge outcomes and discharge location among ICU-admitted patients diagnosed with viral hepatitis using machine learning: a retrospective cohort study of the MIMIC-IV database
BACKGROUND: Hepatitis, a disease characterised by inflammation of the liver, is a leading global health challenge that contributes to over 1.3 million deaths annually, with hepatitis B and C accounting for many of these fatalities. Intensive care unit (ICU) management of patients is particularly challenging due to the complex clinical care and resource demands. Despite advancements in ICU predictive analytics, limited research has specifically addressed hepatitis patients, creating a gap in optimising care for this population. METHODS: This study focuses on predicting ICU length of stay (LoS), hospital discharge outcomes and discharge location for ICU-admitted viral hepatitis patients using a comparative assessment of machine learning (ML) models. Leveraging data from the Medical Information Mart for Intensive Care-IV database, which includes around 94 500 ICU patient records, this study uses sociodemographic details, clinical characteristics and resource utilisation metrics to develop predictive models such as Random Forest, Logistic Regression, Gradient Boosting Machines and Generalised Additive Model with Negative Binomial Regression. RESULTS: Among 3875 ICU-admitted hepatitis patients, Random Forest classification outperformed Logistic Regression in predicting discharge outcomes, achieving higher accuracy (0.87 vs 0.82) and greater discriminative ability (area under the receiver operating characteristic curve 0.95 vs 0.89). For ICU LoS prediction, Random Forest regression applied to log-transformed LoS demonstrated strong performance (R² up to 0.82), while the generalised additive model with negative binomial distribution explained approximately 76% of LoS variance. Prediction of discharge location yielded moderate performance across Gradient Boosting and multinomial logistic regression models (accuracy 0.55 and 0.56), reflecting challenges associated with multi-class imbalance. Variable importance analyses across ML models consistently identified medication counts, procedure counts, comorbidity burden, age, race and total LoS as the most influential predictors of discharge outcomes and discharge location. CONCLUSIONS: This study demonstrates the value of ML models for predicting clinical outcomes for hepatitis patients, including ICU LOS and hospital discharge status. The results underscore the influence of factors like race and age, revealing disparities that must be addressed in predictive care strategies. While the models show promise, challenges such as variability in prolonged stays and limited multiclass prediction accuracy point to the need for ongoing refinement and research.
11 June 2026
Read appraisal →JMIR formative research
User Perspectives on a Clinical Decision Tool to Support Individualized Exercise Prescriptions for Breast Cancer Survivors Not Meeting Exercise Guidelines: Cross-Sectional Survey
BACKGROUND: More than 80% of breast cancer survivors do not meet the recommended levels of exercise, and <50% of health care providers promote exercise as part of survivorship care. Patient-provider communication may enhance exercise engagement by increasing patients' understanding of exercise benefits and linking patients to resources, such as rehabilitation and exercise programs. OBJECTIVE: This study aimed to explore perspectives on a novel clinical decision tool designed to support individualized exercise discussions and prescriptions among breast cancer survivors who do not meet exercise guidelines and health care providers who primarily treat such survivors. METHODS: We conducted a cross-sectional online survey among US breast cancer survivors and health care providers. Participants were (1) female breast cancer survivors aged ≥35 years engaging in ≤150 minutes/week of moderate-intensity aerobic exercise or ≤2 days/week of muscle-strengthening exercise and (2) health care providers who had cared for breast cancer survivors within the past 12 months and reported below-average guideline adherence among their patients. Respondents reviewed a paper draft of a web-based clinical decision prototype tool for supporting individualized exercise discussions and prescriptions based on patients' demographic, clinical, and contextual characteristics. We assessed perceived usefulness, potential uses (eg, counseling), preferred timing of access within clinical encounters, and preferences for tool characteristics (inputs/outputs). RESULTS: The analytic sample comprised 26 breast cancer survivors and 69 health care providers. The survivors' median age was 48 (IQR 37-65) years. Providers included patient navigators/social workers/nurses (29/69, 42.0%), breast oncologists (13/69, 18.8%), and occupational/physical therapists (12/69, 17.4%). The majority of providers (62/69, 89.9%, 95% CI 80.2%-95.8%) and survivors (23/26, 88.5%, 95% CI 69.8%-97.6%) reported that they would find the tool useful. Similarly, 85.5% of providers (59/69, 95% CI 75.0%-92.8%) and 84.6% of survivors (22/26, 95% CI 65.1%-95.6%) reported that the tool would increase their confidence to discuss exercise in a clinical setting. Both groups preferred that survivors access the tool with staff after a medical appointment (survivors: 20/26, 76.9%, 95% CI 56.4%-91.0%; providers: 58/67, 86.6%, 95% CI 76.0%-93.7%). Both groups also endorsed treatment history and readiness to exercise to consider as key inputs and improved quality of life and reduced treatment-related side effects as exercise benefits to communicate as tool outputs. CONCLUSIONS: The prototype tool concept was well received, with high endorsement of individual characteristics to consider and clinical benefits of exercise to communicate. Findings will inform refinement of the tool and future implementation testing in an understudied population of breast cancer survivors.
11 June 2026
Read appraisal →Journal of medical Internet research
Using Ultra-Abridged Individual Difference Scales for Personalization in Digital Mental Health to Improve Uptake, Engagement, and Experiences: Three-Tiered Decision Framework for Scale Shortening
Given the diversity of human characteristics and experiences, personalization in nudges, messages, choice presentations, interventions, and overall product design has been increasingly adopted in digital health to promote engagement. Past studies on moderators and personalization in digital health and mental health services generally focused on demographic and symptom variables, with generally inconsistent findings or null findings. Cognitive, motivational, and decisional psychological attributes are largely overlooked. Psychology often uses long self-report scales to measure various psychological attributes. Although they are useful in tapping into individuals' psychological profiles, when applied in real-life, everyday settings to assess individual differences, people are most likely unwilling to complete them. With the pressing need to personalize digital health platforms to enhance uptake, retention, and engagement, ultrashort versions of these psychological scales may be considered to allow assessment of multiple attributes at the same time. Scale shortening can be achieved through regression analyses of each item, factor analyses, item response theory, ant colony optimization, and machine learning methods, with each method having advantages, disadvantages, and conditions required to make it suitable. To illustrate, we provided examples of regression analyses of each item and factor analyses, with potential implications for personalizing narrative versus research-based messages in digital mental health contexts. We present a 3-tiered decision framework for scale shortening method selection depending on goals and possible constraints, with guidelines on validation methods for ultrashort scales. Moving forward, more validation studies and field studies in digital health platforms are needed to evaluate the ecological validity, reliability, and generalizability of these methods, bearing in mind the limitations and conditions where such shortening methods may not work well. Researchers may compare the effectiveness and limitations of personalization using ultrashort scales with other commonly adopted personalization methods (eg, based on longer scales, behavioral data, and large language models). Ethical concerns need to be considered and mitigated carefully, respecting diverse preferences, informed choices, and the privacy of service users. Our viewpoint piece is primarily intended for digital mental health researchers and practitioners, but may also be informative for the fields of digital health and medicine as well as personalization (eg, personalized health care, personalized nudging, and message matching) more broadly, given the common goal of boosting uptake and engagement as well as improving service users' experiences.
10 June 2026
Read appraisal →International journal of biometeorology
Integrating google trends and hybrid statistical-machine learning models for dengue surveillance in an inland vietnamese province: a 94-month evaluation (2013-2021) with search-signal anomaly screening
Dengue transmission in inland Southeast Asia shows strong seasonality and short-term surges that challenge timely public-health response. We assessed whether province-level Google search activity can support short-horizon dengue monitoring in Dong Nai, Vietnam. Monthly reported dengue cases from July 2013 to April 2021 were aligned with the Google Trends index (GTI) for "Sốt xuất huyết". Models were evaluated using leakage-free rolling-origin expanding-window cross-validation with 3-month validation blocks. We compared a negative binomial (NB) autoregressive baseline, NB models incorporating GTI, tree-based machine-learning models using the same covariates, and a simple NB-random-forest ensemble. Predictive performance was assessed using root mean squared error (RMSE), mean absolute error (MAE), R2, and discrimination for 95th-percentile outbreak exceedance. Cross-correlation peaked at lag 0, indicating that GTI functioned primarily as a contemporaneous nowcasting signal rather than a long-lead predictor. NB models incorporating GTI showed competitive calibration, while boosted tree models achieved the lowest point-estimate errors; XGBoost achieved RMSE ≈ 40.95 and R2 ≈ 0.861 compared with RMSE ≈ 45.23 and R2 ≈ 0.823 for NB with GTI. Adding GTI to the autoregressive NB baseline reduced RMSE by 11.77%, but the 95% confidence interval crossed zero, indicating modest and statistically non-significant incremental gain. Search-signal anomaly screening suggested that GTI spikes largely coincided with epidemic peaks and did not clearly inflate out-of-sample errors. Overall, GTI provides operational value for dengue nowcasting and alert triage in inland provinces, while future systems should integrate climate, vector, mobility, and media-monitoring data.
9 June 2026
Read appraisal →PloS one
Reinforcement learning for policymaking in epidemic control: A scoping review
BACKGROUND: Managing an epidemic demands policies that respond at the pace of the outbreak. Conventional rule‑based interventions struggle to keep up, prompting interest in reinforcement learning (RL) for designing non‑pharmaceutical interventions (NPIs). However, current evidence is fragmented across diverse models and reporting styles. OBJECTIVES: To systematically map how RL is applied for epidemic NPI design, describe modeling choices, algorithm architectures, evaluation practices, and identify trends and research gaps. METHODS: Peer-reviewed studies (2014-2025, English) that applied deep RL to select NPIs were retrieved from IEEE Xplore, ACM Digital Library, ScienceDirect, and Scopus, searched on December 23, 2025. Reference list scanning supplemented database results. Predefined data items (bibliographic details, epidemic and RL model characteristics, experiments, validation methods, outcomes) were charted and summarized descriptively. RESULTS: Of 512 retrieved records, 10 met the inclusion criteria, and three additional studies were identified via reference-list scanning, yielding 13. Five employed value‑based methods, four policy‑gradient, and four hybrid; one study additionally incorporated model-based planning. Six simulations relied on compartmental models, six on agent‑based models, and one on a hybrid model. Action spaces were predominantly discrete restriction levels. Five studies incorporated sequence-modeling techniques to include temporal context into a state space. Eleven studies designed reward functions as a trade-off between pandemic severity and socio-economic cost. According to the reviewed studies, RL policies across various settings outperform heuristic, rule-based, and historical baselines in reducing infections, deaths, or lockdown duration while limiting economic loss. CONCLUSIONS: RL shows promise for adaptive epidemic control. Comparison is hampered by simplified economic costs, inconsistent calibration rigor, varied evaluation metrics, and limited uncertainty or policy robustness analysis. Future work should establish common benchmark environments and reporting standards, incorporate empirically grounded economic and behavioral models, adopt uncertainty-aware and probabilistic RL, develop more sophisticated control spaces, investigate more advanced algorithms, and validate learned policies prospectively to enable real-world deployment.
9 June 2026
Read appraisal →Proceedings of the National Academy of Sciences of the United States of America
Advancing AI negotiations: A large-scale autonomous negotiation competition.
We conducted an international AI negotiation competition in which participants designed and refined prompts for AI negotiation agents. We then facilitated over 180,000 negotiations between these agents across multiple scenarios with diverse characteristics and objectives. Our findings revealed that principles from human negotiation theory remain crucial even in AI-AI contexts. Surprisingly, warmth-a traditionally human relationship-building trait-was consistently associated with superior outcomes across all key performance metrics. Dominant agents, meanwhile, were especially effective at claiming value. Our analysis also revealed unique dynamics in AI-AI negotiations not fully explained by negotiation theory, including AI-specific technical strategies like chain-of-thought reasoning and prompt injection. When we applied natural language processing methods to the full transcripts of all negotiations, we found positivity, gratitude, and question-asking (associated with warmth) were strongly associated with reaching deals as well as objective and subjective value, whereas conversation lengths (associated with dominance) were strongly associated with impasses. The results suggest the need to establish a new theory of AI negotiation, which integrates classic negotiation theory with AI-specific negotiation theories to better understand autonomous negotiations and optimize agent performance.
7 June 2026
Read appraisal →PloS one
The application of large language models in bariatric surgery: A scoping review
BACKGROUND: Exploratory applications of large language models within the specialized field of metabolic and bariatric surgery have begun to emerge. Nevertheless, existing research remains fragmented, lacking comprehensive integration. OBJECTIVE: To conduct a scoping review of studies on the application of large language models in the field of metabolic and bariatric surgery, aiming to provide a reference for clinical practice and future research. METHODS: This scoping review adhered to the Joanna Briggs Institute methodological framework and followed the preferred reporting items for systematic reviews and meta-Analyses extension for scoping reviews (PRISMA-ScR) guidelines.PubMed, Web of Science, The Cochrane Library, Embase, CINAHL, CNKI, Wanfang, and VIP databases were searched for relevant studies, with the search timeframe from database inception to November 2025. The included literature was summarized and analyzed. RESULTS: A total of 21 English-language studies were included. LLMs were primarily applied in scenarios such as patient education and information consultation, clinical decision support, and professional knowledge assessment. While LLMs performed well in information-provision tasks, they showed low consistency with expert opinions in complex clinical tasks such as individualized surgical recommendations. Performance varied across different models, with GPT-4 generally demonstrating superior performance, and domain-specific models showing professional potential. Current research still faces challenges regarding information accuracy, readability, and clinical applicability. CONCLUSION: Large language models hold auxiliary potential in the field of metabolic and bariatric surgery, particularly for knowledge dissemination and patient education. However, their reliability in complex clinical decision-making remains limited. Future efforts should focus on conducting high-quality studies, advancing model specialization and standardized evaluation, and exploring safe and effective human-AI collaboration models.
7 June 2026
Read appraisal →BMJ health & care informatics
Routine data are not 'dirty' data: how to approach using routine data for research.
The rise of electronic patient records has led to an abundance of routine healthcare data, and the development of NHS Secure Data Environments (SDEs) is transforming their potential for research. However, routine data differ fundamentally from research data; they are generated to support care processes rather than answer predefined research questions. This creates challenges when they are analysed without understanding the context in which they were produced.Using real examples from practice, this paper illustrates the variety of ways routine data can be shaped and how this leads to misinterpretation if treated as merely 'dirty' research data. These examples underpin practical strategies for researchers, including analysis at or by source and working closely with local data providers to understand how a dataset was created.We argue that SDEs must foster strong links between data users and creators to preserve data provenance and support accurate interpretation. By approaching routine data as meaningful records of healthcare processes-rather than flawed research datasets-researchers can fully realise their potential to drive healthcare improvement.
1 June 2026
Read appraisal →JMIR research protocols
Creation of an mHealth Infrastructure to Support the Development and Delivery of mHealth Interventions: Protocol for Demonstration Projects Addressing Smoking Cessation in Cancer Care
BACKGROUND: Cancer remains a leading cause of morbidity worldwide. To reduce this burden, scalable, effective approaches are needed to address modifiable risk factors for cancer and support behavioral self-management. With smartphone ownership now nearly ubiquitous, mobile health (mHealth) interventions offer a powerful means to extend the reach, accessibility, and sustainability of evidence-based treatments for a variety of modifiable risk factors (eg, excessive alcohol use, physical inactivity, poor diet, and smoking). Moreover, the flexibility of mHealth platforms enables efficient delivery of novel interventions, supports innovative study designs, and facilitates real-time data collection to advance public health research. OBJECTIVE: Despite the great potential of mHealth interventions, developing high-quality mHealth tools is complex, time-consuming, and resource-intensive. To address these challenges, we are developing a coordinated, accessible, research-grade infrastructure for mHealth app development, testing, and dissemination. METHODS: The mHealth Florida infrastructure (mFLi) will provide a comprehensive, low-code software platform that enables researchers to build apps compatible with major mobile operating systems, namely, Apple iOS and Google Android. Through a modular interface, users will select from a menu of prebuilt features to tailor functionality to specific study needs. The platform will include 3 integrated environments (development, testing, and production), allowing researchers to prototype, evaluate, and deploy mHealth interventions. This infrastructure will be developed and maintained by a multidisciplinary team, ensuring that the platform is technically robust and usable and adheres to institutional and regulatory standards. To demonstrate the platform's functionality, utility, and adaptability, a multisite study comprising three initial projects focused on smoking cessation among patients with cancer is being conducted: (1) participant screening and enrollment, (2) randomization and treatment delivery, and (3) data processing using machine learning methods with on-device and cloud-based approaches. RESULTS: This study was funded in May 2023, and ethics approval was obtained from all involved sites' institutional review boards between February 2024 and October 2025. Recruitment began in March 2025 and enrollment is ongoing. As of January 2026, 41% (37/90) of the target sample have been enrolled and 21% (19/90) have completed their 6-month assessment. Data collection will be completed once the final participant completes their 6-month assessment (expected May 2027), with analyses commencing thereafter. Study findings are anticipated to be published in a peer-reviewed journal in 2027. CONCLUSIONS: Collectively, these projects will illustrate how mFLi can streamline app development, facilitate rapid translation of research into practice, and reduce barriers for researchers and developers. Ultimately, mFLi is designed to accelerate innovation in mHealth research, enhance access to behavioral interventions, and improve health outcomes among diverse populations. TRIAL REGISTRATION: ClinicalTrials.gov NCT06909357; https://clinicaltrials.gov/study/NCT06909357. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): PRR1-10.2196/92288.
1 June 2026
Read appraisal →Critical care explorations
Linking the Virtual Pediatric System and Pediatric Hospital Information System Databases to Create Novel PICU Research Datasets
BACKGROUND: Lack of a robust, national PICU database hinders epidemiologic and quasi-experimental research of PICU care. Combining the virtual pediatric system (VPS) and pediatric health information system (PHIS) databases would overcome this data gap. However, limits on the use of variables containing protected health information (PHI) have prevented prior large-scale VPS-PHIS linkage. We sought to link VPS and PHIS records from 31 hospitals without using PHI. METHODS: We performed a database linkage study utilizing a sequential direct + probabilistic linkage approach with records from 31 hospitals contributing to VPS and PHIS for patients discharged between January 1, 2017 and December 31, 2022. Using directly linked records as a reference, we assessed the accuracy of probabilistically linked pairs with data from four hospitals by describing the percent of false-negative and false-positive links. We also compared demographic and clinical variables between linked and unlinked records using Cohen's Effect size (CES). RESULTS: We obtained 337,836 VPS and 3,515,595 PHIS records without PHI. In total, we linked 319,955 of 337,836 (94.7%) VPS records. Direct linkage contributed 274,345 of 319,955 (85.7%) of all record linkages. Probabilistic linkage contributed 45,610 of 319,955 (14.3%) of all record linkages. Assessment of the probabilistic linkage found 0.21% false-negative links and no false-positive links. We found negligible imbalance by CES for gender (% female in linked vs. unlinked: 44.8% vs. 43.6%, CES 0.024), age in years (median [interquartile range (IQR)]: 5.1 [1.21-13.22] vs. 5.53 [1.53-13.38], CES 0.054), Pediatric Risk of Mortality 3 score (median [IQR]: 0.44 [0.3-0.9] vs. 0.37 [0.29-1.02], CES 0.096), or days in PICU (median [IQR]: 1.55 [0.8-3.47] vs. 1.33 [0.79-3.45], CES 0.031). CONCLUSIONS: Using a combined direct and probabilistic linkage approach, we created a highly accurate linkage VPS and PHIS database records from 31 hospitals without using PHI. This combined database can be used to create novel datasets for PICU research.
31 May 2026
Read appraisal →Medicine
Machine learning-based identification of multidimensional predictors of quality of life in individuals with multiple sclerosis
This study aimed to identify independent predictors of quality of life in patients with multiple sclerosis (MS) using machine learning approaches. One hundred and one individuals diagnosed with MS were included in this cross-sectional study. Quality of life was assessed using the MS Quality of Life-54 (MSQoL-54). Demographic variables, clinical characteristics, disability level (Expanded Disability Status Scale [EDSS]), fatigue severity, sleep quality (Pittsburgh Sleep Quality Index), depression level (Beck Depression Inventory), and functional mobility (Timed Up and Go test) were evaluated. Multiple machine learning regression models, including Linear Regression, Lasso, Elastic Net, Support Vector Machines, Random Forest, and XGBoost, were developed and compared. Five-times repeated five-fold cross-validation was applied for internal validation. Model performance was evaluated using root mean squared error (RMSE), mean absolute error (MAE), and coefficient of determination (R2). Among all models, Lasso regression demonstrated the best predictive performance (RMSE = 5.02, MAE = 4.04, R2 = 0.86). Disability level (EDSS) emerged as the strongest negative predictor of quality of life, followed by fatigue severity, sleep quality, and impaired functional mobility. Although disease duration was not significantly correlated with quality of life in univariate analyses, it showed a positive contribution in the multivariable model. Depression showed a strong bivariate association but had a relatively lower weight in the final predictive model. Quality of life in individuals with MS is strongly influenced by disability, fatigue, sleep quality, and functional mobility. Machine learning approaches provide an effective and interpretable framework for identifying key determinants of quality of life and may support personalized rehabilitation and clinical decision-making in MS.
31 May 2026
Read appraisal →European radiology experimental
Towards an accessible, centralised, searchable database for AI courses in Europe: the Artificial Intelligence in Medical Imaging and Radiation Oncology Education (AIMIROE) project.
OBJECTIVE: Artificial intelligence (AI) is transforming medical imaging and radiation oncology, yet limited understanding and access to education hinder adoption. This study, led by the European Society of Medical Imaging Informatics (EuSoMII) in collaboration with the European Federation of Radiographer Societies (EFRS), aimed to create an accessible, centralised, searchable database including all AI courses in Europe. MATERIALS AND METHODS: An electronic survey was developed to collect data on European AI course characteristics, such as format, delivery, content, target audience and European Qualifications Framework (EQF) level. This was disseminated via purposive sampling through social media and mailing lists of the EuSoMII and the EFRS between September 2024 and January 2025. Quantitative data were analysed using descriptive statistics and visual representations using Python Seaborn and Geopandas. RESULTS: This study identified 29 AI courses in Europe. Of them, 53.6% were offered by universities. Courses targeted radiographers (59%), medical physicists (52%), and radiologists (41%), mainly at EQF level 7 (44.4%). Most courses were standalone (65.6%) and online (55.1%), while 41.3% were free of charge. English was the primary language of delivery (79%). CONCLUSIONS: Different AI courses across Europe offer some entry-level knowledge but are often short in duration. Expanding formats, building practical competencies, providing multilingual access, and European-wide reach are essential for meaningful, practical, and equitable AI integration. RELEVANCE STATEMENT: With the scaling-up of AI adoption in medical imaging and radiation oncology, there is a variety of AI education provisions currently available. Accessing these options via an open, centralised, regularly updated database enables people to make an informed decision about their training and practise safely and meaningfully. KEY POINTS: We identified 29 different AI European courses varying in language, content, and delivery. Many clinical practitioners and researchers are unaware of these resources. We need a centralised database for customising AI learning choices and guiding future course design.
31 May 2026
Read appraisal →JMIR medical informatics
The Regional Implementation of an Electronic Health Record-Integrated Ambient Scribe in Primary and Secondary Care in England: Real-Time Qualitative Evaluation
BACKGROUND: There is significant potential for ambient scribe technology to enhance health care productivity, with a growing range of applications being developed and implemented internationally. Strong organizational drivers to improve efficiency, coupled with the technology's potential to help address clinician burnout, are accelerating interest and adoption. However, limited attention has been paid to the integration of such systems within electronic health records and unintended consequences, as stakeholders navigate their implementation and integration into clinical practice. OBJECTIVE: This study therefore aimed to explore the processes involved in implementing and adopting ambient scribe technology across diverse health care settings. METHODS: We conducted a real-time, longitudinal qualitative evaluation of a pilot implementation of an integrated ambient scribe system across National Health Service primary care and secondary hospital settings within a care system in the Midlands region of England. Data collection involved in-depth one-to-one interviews conducted in several phases: an initial scoping study to identify key interests and stakeholders for inclusion, followed by an implementation study with participants involved in the pilot. We also conducted 16 hours of nonparticipant observations of consultations. The implementation study gathered data at 2 time points, before implementation and 3 to 4 weeks after, to capture experiences, changes, and emerging impacts over time. RESULTS: We collected data from 45 individuals. Use cases varied across settings and shaped how ambient scribe systems were deployed. Differences between general practice and secondary care in documentation purpose, format, and workflow created challenges for developing and validating templates, though the technology showed flexibility across contexts. Automated note-taking often improved patient interaction but required clinicians to adjust how they spoke to ensure the technology captured their reasoning. Outputs were sometimes generic, reinforcing defensive documentation and reducing personal tone or contextual recall. Integration pathways carried distinct trade-offs: stand-alone systems (which were used by many stakeholders in our study) were easier to adopt but offered limited long-term benefits, while integrated systems required greater effort and standardization, yet promised improved efficiency and safety. CONCLUSIONS: The ambient scribe market remains immature and volatile, creating strategic uncertainty for health systems. Careful procurement approaches are needed to balance the risks and benefits of integrated versus stand-alone systems, while aligning user demand with organizational needs for integration.
31 May 2026
Read appraisal →African journal of reproductive health
The role of information technology in promoting sexual and reproductive health awareness among young people in Lahore: A comparative study with Shanghai.
Information technology (IT) has transformed access to health information, yet its impact on sexual and reproductive health (SRH) awareness varies across contexts. This study investigates IT's influence on youth SRH awareness in Lahore, Pakistan, using primary data, and provides a comparative analysis with secondary data from Shanghai, China. Using a cross-sectional design in Lahore (n=250), structured questionnaires, PLS-SEM, logistic regression, and ANOVA assessed relationships between IT use, digital health literacy, trust in digital sources, cultural barriers, and awareness outcomes. Results indicate IT use (β=0.42), digital health literacy (β=0.38), and trust (β=0.26) significantly enhance awareness in Lahore, while cultural barriers exert a strong negative influence (β=-0.31). Significant mean differences (p<0.001) were found between cities. Shanghai demonstrated higher digital literacy, trust, awareness, and use of structured platforms, while Lahore exhibited substantially higher cultural barriers. Model comparisons showed stronger predictive pathways in Shanghai (R²=0.74) than Lahore (R²=0.68), reflecting advantages of institutionalized sex education and government-certified digital health applications. Overall, while IT holds considerable potential, its effectiveness is shaped by cultural norms, digital infrastructure, and institutional trust. The findings underscore Pakistan's need to develop culturally appropriate, reliable, and institutionally supported digital health platforms to improve youth access to accurate SRH information. La technologie de l'information (TI) a transformé l'accès à l'information sanitaire, mais son impact sur la sensibilisation à la santé sexuelle et reproductive (SSR) varie selon les contextes. Cette étude examine l'influence des TI sur la sensibilisation des jeunes à la SSR à Lahore, au Pakistan, avec une analyse comparative de Shanghai, en Chine. En utilisant un design transversal à Lahore (n=250), des questionnaires structurés, la PLS-SEM, la régression logistique et l'ANOVA ont évalué les relations entre l'utilisation des TI, la littératie numérique en santé, la confiance dans les sources numériques, les barrières culturelles et les résultats de sensibilisation. Les résultats indiquent que l'utilisation des TI (β=0,42), la littératie numérique en santé (β=0,38) et la confiance (β=0,26) améliorent significativement la sensibilisation à Lahore, tandis que les barrières culturelles exercent une forte influence négative (β=-0,31). Des différences moyennes significatives (p<0,001) ont été constatées entre les villes. Shanghai a démontré une littératie numérique, une confiance, une sensibilisation et une utilisation de plateformes structurées plus élevées, tandis que Lahore présentait des barrières culturelles substantiellement plus fortes. Les comparaisons de modèles ont montré des voies prédictives plus solides à Shanghai (R²=0,74) qu'à Lahore (R²=0,68), reflétant les avantages de l'éducation sexuelle institutionnalisée et des applications de santé numériques certifiées par le gouvernement. Globalement, bien que les TI aient un potentiel considérable, leur efficacité est façonnée par les normes culturelles, l'infrastructure numérique et la confiance institutionnelle. Les conclusions soulignent la nécessité pour le Pakistan de développer des plateformes de santé numériques culturellement adaptées, fiables et soutenues institutionnellement pour améliorer l'accès des jeunes à des informations précises sur la SSR.
31 May 2026
Read appraisal →JMIR infodemiology
Early Depression Detection in Social Media: Monitoring of Individual Nighttime Dynamics and Large Language Model Analysis
BACKGROUND: Depression has become a major global public health challenge, and early intervention is critical for improving patient outcomes. Current depression detection techniques based on social media data (traditional risk detection) rely heavily on users' complete historical information, which cannot meet the timeliness requirements of early intervention. This underscores the need for early risk detection (ERD) methods emphasizing early-stage, real-time warning. However, existing ERD studies have notable limitations such as (1) they overlook temporal activity patterns hidden in posting time stamps, missing vital warning signals; and (2) they depend on static templates or resource-intensive sequence models, resulting in limited interpretability and inefficient use of early data, ultimately constraining their clinical applicability for early intervention. OBJECTIVE: This study aims to develop an efficient, reliable, and interpretable ERD model. The core objectives are to extract temporal activity patterns features from posting time stamps, thereby enriching feature dimensions for risk detection; to leverage large language models (LLMs) for improved text filtering precision and depression-related factor analysis; and to achieve accurate early detection of depression to support clinical intervention. METHODS: We propose the Monitoring of Individual Nighttime Dynamics (MIND) and LLM analysis model, which integrates two key innovations: (1) circadian activity dynamics: posting time stamps are transformed into temporal activity patterns, analyzing fluctuations in posting frequency and timing to derive sleep-related features, thereby compensating for the limitations of text-only approaches, and (2) LLM depression profiler: LLMs are used for dynamic text filtering, automatically removing irrelevant noise and focusing on potential depression-related cues. Based on LLM semantic understanding, latent depression risk factors are identified, enhancing interpretability for clinical treatment and robustness to noise. RESULTS: Experiments on the eRisk2017 (data source: Reddit [Reddit Inc]) benchmark dataset demonstrated that MIND significantly outperformed existing baseline models in early detection sensitivity, specificity, and accuracy. By combining sleep-related features with text analysis, the model achieved interpretable, traceable predictions that can support clinical treatment. ALL relevant experimental code is publicly available. CONCLUSIONS: The MIND model combines temporal activity pattern features with LLM-based text analysis, addressing the challenges of poor interpretability and inefficient use of early-stage data in existing ERD methods. It significantly enhances early detection performance, offering a new paradigm for applying social media data in ERD, thereby enabling earlier intervention and reducing the public health burden of depression.
31 May 2026
Read appraisal →BMJ open
Impacts of human and digital scribes on improving physician wellness and reducing administrative burden: a scoping review protocol
INTRODUCTION: The rise in administrative burden on physicians is becoming increasingly recognised as a significant contributor to burnout and job dissatisfaction among primary care practitioners. Human scribes (HS) and digital/artificial intelligence scribes (DS) have emerged as potential tools to reduce clerical workload and improve physician well-being. There has been extensive research conducted on HS, and DS show promise but require validation across diverse healthcare settings. This scoping review aims to assess the effects of HS and DS in primary care settings to evaluate their impact on administrative work, job satisfaction, burnout and clinical operational efficiency. METHODS AND ANALYSIS: This review will follow the Arksey and O'Malley framework with enhancements from Levac et al, along with Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) reporting guidelines. A comprehensive search strategy was developed, including Ovid MEDLINE, Ovid Embase and Scopus databases in June 2025 with an academic health sciences librarian. Database searches will be conducted between March and June 2026. The inclusion criteria consist of empirical research on licensed physicians in primary care settings that implement HS or DS systems and measure physician-related outcomes. Two researchers will review and extract data independently using Covidence. The results will be analysed through thematic methods followed by descriptive summarisation. ETHICS AND DISSEMINATION: This review conducts an analysis of existing publications without needing ethical clearance. Findings will be disseminated through peer-reviewed publications and presented at academic conferences where possible. The research team will make supporting materials accessible via the Open Science Framework. TRIAL REGISTRATION NUMBER: https://doi.org/10.17605/OSF.IO/T5BHY.
31 May 2026
Read appraisal →Purinergic signalling
Turning failure into success: how artificial intelligence can help personalize therapies and re-use patient data.
Despite robust preclinical evidence, many clinical trials, including several that targeted the purinergic system, fail to demonstrate efficacy in humans. Failure may stem from inability to accurately identify patient subgroups responding similarly to treatments. Here, we explore the potential of artificial intelligence to revolutionize how we group and classify patients in clinical studies. We introduce a new framework using Large Language Models-generated embeddings of detailed patient data, to create a semantic-aware latent space, enabling us to identify truly meaningful patients' clusters. Large Language Models can provide explainable groupings, giving clear reasons why certain patients respond similarly to treatments. We present an example of successful application of this approach through the re-analysis of the AMARANTH clinical trial (NCT02245737, involving ~ 2200 patients and completed in 2018) testing Lanabecestat, a BACE1 inhibitor decreasing β-amyloid production in Alzheimer's disease, for which traditional analysis reported no efficacy. As in the original trial, our simulation showed no overall benefit. However, re-analysis per patients' clusters and subjects' re-stratification by semantic similarities (shared symptom profiles, progression patterns) identified a patients' subgroup in one of the clusters showing Lanabecestat-associated slower disease worsening, thus succeeding where the full trial had failed. By making a new therapy available to at least a subset of patients with a defined disease, this new approach may help maximize the return on drug development and reduce the burden on healthcare. Moreover, it will significantly improve the precision, efficiency and interpretability of clinical trials, paving the way for a new era of personalized medical treatments.
31 May 2026
Read appraisal →JMIR formative research
Machine Learning-Based Prediction Model for 30-Day Emergency Department Revisits in a Medically Underserved Tertiary Hospital: Formative Retrospective Cohort Study
BACKGROUND: Emergency department (ED) revisits are critical quality indicators, particularly in medically underserved areas, where traditional prediction tools show limited performance. Machine learning (ML) approaches may offer improved predictive performance for identifying high-risk patients. OBJECTIVE: This formative study aimed to develop and validate an ML-based model for predicting 30-day ED revisits using electronic health records from a tertiary hospital serving a medically underserved area in South Korea and to evaluate its clinical utility through interpretability analysis and risk stratification. METHODS: This retrospective cohort study analyzed 36,230 adult patients visiting the Gangneung Asan Hospital ED in 2023. We developed and compared 3 ML models (extreme gradient boosting [XGBoost], random forest, and ElasticNet) using electronic health records. Model interpretability was ensured through Shapley additive explanations (SHAP) analysis, and clinical utility was evaluated through 5-tier risk stratification. RESULTS: Among 36,230 patients, 798 (2.2%) revisited within 30 days. XGBoost achieved superior performance with an area under the receiver operating characteristic curve of 0.90 (95% CI 0.88-0.92), a sensitivity of 0.94, and a specificity of 0.69. The SHAP analysis identified ED length of stay, oxygen saturation, systolic blood pressure, computed tomography performance, antibiotic use, and liver disease as key predictors. Risk stratification demonstrated a 25-fold difference in the actual revisit rates between the lowest (152/8450, 1.8%) and the highest (686/1500, 45.7%) risk groups. CONCLUSIONS: The XGBoost model demonstrated excellent predictive performance with high interpretability for 30-day ED revisit predictions. The implementation of this model could enable risk-stratified interventions and more efficient resource allocation in medically underserved settings, potentially reducing unnecessary revisits and improving patient outcomes. This formative study establishes feasibility and provides a foundation for future multicenter validation studies in similar medically underserved settings.
30 May 2026
Read appraisal →African journal of reproductive health
Artificial intelligence-enabled digital reproductive health education: corpus-based strategies for strengthening Spanish health communication capacity
This study examines how corpus-grounded artificial intelligence (AI) can strengthen Spanish reproductive health communication capacity within China's digital health ecosystem. A mixed-methods design was employed, combining corpus linguistics, AI-assisted message generation, expert-informed evaluation, and quantitative user assessment. A domain-specific Spanish reproductive health corpus was constructed from 3,052 documents, including clinical guidelines, patient education materials, and FAQs/user queries, yielding 1,065,110 tokens for linguistic analysis and AI grounding. Corpus-derived readability benchmarks, lexical simplification rules, and discourse patterns were integrated into an AI content generation pipeline to produce reproductive health messages, which were then compared with non-corpus-grounded AI outputs. The user evaluation phase was conducted among 240 Spanish-speaking or Spanish-proficient adults in selected Chinese metropolitan cities. Data were collected through a structured questionnaire measuring Corpus-Grounded AI Generation, Spanish Communication Quality, and Health Communication Capacity, and were analyzed using reliability testing, correlation analysis, and structural equation modeling. The findings showed that corpus-grounded AI significantly improved Spanish communication quality, while communication quality had the strongest effect on users' comprehension, confidence, and help-seeking intention. Mediation analysis further demonstrated that Spanish communication quality significantly mediated the relationship between corpus-grounded AI generation and health communication capacity. The study concludes that linguistically informed AI design can enhance the clarity, accessibility, and effectiveness of reproductive health education and offers a practical framework for multilingual digital health communication in sensitive healthcare contexts. Cette étude visait à examiner le rôle de l'IA guidée par la linguistique de corpus dans l'amélioration de la qualité de la communication en espagnol sur la santé reproductive au sein de l'écosystème numérique chinois. En utilisant une approche à méthodes mixtes, la recherche a combiné l'analyse de corpus de textes espagnols sur la santé reproductive, la génération de contenu basée sur l'IA et l'évaluation par les utilisateurs. Un corpus espagnol spécifique au domaine de la santé reproductive a été construit et intégré dans un pipeline de génération d'IA ancré dans le corpus pour optimiser la lisibilité, l'exactitude et la sensibilité culturelle. La qualité de la communication et la capacité de communication en santé ont été mesurées à l'aide d'une analyse de fiabilité, de tests de corrélation et de modélisation par équations structurelles. Les résultats ont révélé que l'IA ancrée dans le corpus améliore significativement la qualité de la communication en espagnol, ce qui améliore à son tour la compréhension des utilisateurs, l'auto-efficacité et les intentions de recherche d'aide. La qualité de la communication est apparue comme un mécanisme médiateur clé entre la conception de l'IA et les résultats en matière de communication santé. Sur le plan théorique, l'étude établit une base pour intégrer la linguistique de corpus dans les systèmes d'éducation à la santé pilotés par l'IA, tandis que sur le plan pratique, elle propose un cadre évolutif pour les plateformes multilingues d'éducation à la santé reproductive. Ces résultats ont des implications importantes pour les politiques mondiales de santé numérique, la gouvernance de l'IA et les stratégies de communication en santé reproductive.
30 May 2026
Read appraisal →JMIR medical informatics
Blockchain Smart Contracts for Automating Clinical Trials: Systematic Review and Proposed System Architecture
BACKGROUND: Blockchain technologies have revolutionized the financial sector through their ability to generate immutable, cryptographically secure records. Clinical trials and health care data possess several synergies with those of the financial sector, specifically pertaining to the importance of tamper-resistant recording of processes. The evolution of blockchain to autonomously execute tasks contingent upon predefined contractual terms via smart contracts (SCs) allows a dynamic chain of interlinked events to unfold independently and in sequence, with time-stamped records. In recent years, mistrust in clinical trial data has grown significantly. Recording the entire clinical trial lifecycle from application, registration, recruitment to conduct, finance management, statistical analysis, and reporting in an immutable, cryptographically secure ledger with SC execution of trial processes could limit the potential for human intervention and tampering. This would produce a time-stamped record of all events within the trial lifecycle. Leveraging the capabilities of SCs could alleviate recruitment challenges and address ongoing concerns regarding data transparency, ownership, and integrity that currently undermine clinical trial processes. OBJECTIVE: This study aimed to review the existing literature on SC applications in clinical trials and propose a system architecture for using SCs to automate key processes throughout the clinical trial lifecycle. METHODS: A systematic search was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, identifying peer-reviewed studies and open-source repositories pertaining to the implementation of SCs in clinical trials. Data were extracted specific to the stage of the trial lifecycle described, SC architecture, and technical specifications for real-world implementation. Data were synthesized to propose an architecture for automating clinical trial processes within the lifecycle using SCs. RESULTS: A total of 144 records were screened; 10 studies met the inclusion criteria. Most implementations used private Ethereum-based networks (7/10, 70%). Reported applications included automated patient matching (5/10, 50%), consent management with dynamic permissioning (6/10, 60%), protocol enforcement and time-stamped audit logs (9/10, 90%), adverse event reporting (3/10, 30%), and financial or workflow automation (3/10, 30%). SC-based recruitment systems demonstrated rapid matching performance (eg, 6000 simulated patients matched in 2.13 s). However, all included systems were prototypes or simulations, and none were tested in real-world regulatory settings. Scalability, interoperability limitations, regulatory ambiguity (eg, General Data Protection Regulation right-to-erasure conflicts), and high infrastructural complexity were common gaps noted across studies. CONCLUSIONS: Current evidence suggests that SCs can enhance transparency, traceability, and automation throughout the clinical trial lifecycle. However, the literature remains dominated by simulation-based prototypes, primarily Ethereum-dependent architectures, and lacks analyses of cost-effectiveness, governance, and integration with institutional workflows. Future research should evaluate hybrid architectures, develop interoperability standards, and assess regulatory and ethical implications in real deployments.
30 May 2026
Read appraisal →Philosophical transactions. Series A, Mathematical, physical, and engineering sciences
Predictability-Computability-Stability workflow for veridical data science in the age of artificial intelligence
Data science is a pillar of artificial intelligence (AI), which is transforming nearly every domain of human activity, from the social and physical sciences to engineering and medicine. While data-driven findings in AI offer unprecedented power to extract insights and guide decision-making, many are difficult or impossible to replicate. A key reason for this challenge is the uncertainty introduced by the many choices made throughout the data science life cycle (DSLC). Traditional statistical frameworks often fail to account for this uncertainty. The Predictability-Computability-Stability (PCS) framework for veridical (truthful) data science (VDS) offers a principled approach to addressing this challenge throughout the DSLC. This paper presents an updated and streamlined PCS workflow, tailored for practitioners and enhanced with guided use of generative AI (GenAI). We include a running example to display the PCS framework in action and conduct a related case study that showcases the uncertainty in downstream predictions caused by judgement calls in the data cleaning stage. This article is part of the theme issue 'Statistical workflow'.
30 May 2026
Read appraisal →PloS one
Risk assessment of civil aviation cabin safety incidents based on the CNN-LSTM-Attention model
Against the backdrop of the thriving global civil aviation industry, cabin safety, which is a crucial link in civil aviation safety systems, has prompted an increasingly urgent need for intelligent risk management. This study constructs an integrated framework of "deep text mining-intelligent risk assessment" and proposes a hybrid CNN-LSTM-Attention model for cabin safety incident classification. Using 8,280 global cabin abnormal event reports from 2004-2024, the model achieves 95.01% accuracy and an F1 score of 94.17% on the test set, substantially outperforming benchmark approaches. Compared with XGBoost (89.45% accuracy, 88.98% F1), the proposed model improves accuracy by 5.56% and the F1 score by 5.19%, demonstrating superior capability in extracting deep semantic features and identifying key risk patterns. The framework also establishes a "text feature-risk mechanism-hazard level" mapping system, enabling more interpretable risk quantification. The findings provide an effective technical path for intelligent early warning of cabin safety risks and offer methodological support for data-driven decision-making in civil aviation safety management.
30 May 2026
Read appraisal →Current atherosclerosis reports
Patient Engagement Strategies for Implementing Evidence-Based Care in Atherosclerosis
PURPOSE OF THE REVIEW: Despite continuing therapeutic advances in cardiovascular disease, it remains a leading cause of morbidity and mortality in the US. The current challenge is not the development of therapeutics but rather patient engagement and compliance with medical recommendations. Here in this narrative expert review we discuss strategies to improve patient engagement in cardiovascular care. RECENT FINDINGS: There is a varying level of data surrounding different patient education strategies in the care of atherosclerosis. There are three main categories of interventions reviewed with evidence of efficacy: personalized and accessible patient education, digital health interventions, and implementation of personalized medicine via genetic testing. Implementing strategies in the clinic such as personalizededucation at an accessible health literacy level, digital health interventions,and even genetic testing may improve patient engagement and outcomes.
30 May 2026
Read appraisal →PloS one
Transformation of artistic style and innovative design of oriental folk patterns based on AIGC Technology-A case study of Zhuxian town new year paintings from China
In response to the limited cross-domain innovation in the digitalization of traditional oriental folk art, this study takes the New Year pictures of China's Zhuxian Town as a case. It develops a collaborative technical framework of combining the Liblib platform and a LoRA model to explore AI-generated digital re-creation of traditional art. A high-quality dataset was built through a three-stage process of image acquisition, multidimensional screening, and expert review. A three-layer keywords thesaurus was constructed through literature analysis, questionnaire surveys, and semantic clustering. Using a two-stage training strategy combining pre-training and fine-tuning, along with dynamic optimization, the model accurately captures the stylistic features of Zhuxian New Year pictures. The generated outputs integrate traditional aesthetics with modern design and are applied to cultural and graphic creative products. The results demonstrate the effectiveness of the proposed framework for preserving artistic style while enabling cross-domain innovation and offer a practical technical reference for the digital inheritance and modernization of related folk arts.
30 May 2026
Read appraisal →Journal of medical Internet research
Can Digital Tools Fix Bias in Mental Health Triage?
Recent advances in digital technologies hold both promise and potential pitfalls for mental health triage. In this News and Perspectives article, JMIR Correspondent Beth Rush reports on the issue of bias and the role of digital tools in the mental health context.
29 May 2026
Read appraisal →JMIR aging
Developing Customized Personas to Capture Intrinsic Capacity Profiles and Digital Monitoring Intentions in Older Adults: Mixed Methods Study
BACKGROUND: Integrated Care for Older People (ICOPE), focused on monitoring and optimizing the intrinsic capacity (IC) of older adults, is a new model of geriatric care that is currently being accelerated globally. Digital health technologies are recommended for longitudinal IC monitoring to provide precise and timely interventions. However, little is known about the psychological intentions of engaging in digital monitoring of IC according to the profile heterogeneity of IC among older adults. OBJECTIVE: This study aims to map a set of customized personas to capture the profiles of IC and match psychological intentions that support personalized digital IC monitoring. METHODS: An explanatory sequential mixed methods study was conducted at 16 sites in Beijing, China. Older adults aged ≥60 years (n=481) were selected to complete the quantitative survey. Latent profile analysis, descriptive statistics, and logistic regression analyses were performed to cluster subgroups using Mplus (Muthén & Muthén) and SPSS (IBM Corp). A subsample of participants from each profile (n=25) was purposively sampled for semistructured interviews. An inductive-deductive content analysis was used to identify similar attributes and to affirm the personas gradually. A joint statistical and thematic visualization method was used to integrate the customized personas. RESULTS: Three profiles of IC patterns emerged: "multisubdomain decline-IC imbalance group," "multisubdomain moderate-sensory deficit group," and "multisubdomain robust-whole balance group." The distribution of latent profiles was influenced by age, education, monthly per capita household income, self-rated health, and number of chronic diseases, while positively impacting older adults' functional ability. The following customized personas were captured regarding established themes: "affects my mood-anxious evader," characterized by avoidance and anxiety, low digital interest, and perceived social isolation; "capitalize on what comes-accommodative adopter," pragmatically oriented toward disease detection, with moderate digital openness but limited self-efficacy; and "more autonomy-active improver," who exhibited proactive engagement, high digital literacy, and motivation rooted in self-management and social participation. CONCLUSIONS: This study is the first to integrate latent profile analysis with customized qualitative personas to link the heterogeneity of IC with the psychological intentions underlying digital monitoring. The resulting personas model provides an actionable framework for tailoring digital IC monitoring strategies in community-based integrated care. The findings emphasize the need to align monitoring approaches with older adults' IC characteristics, psychological readiness, digital literacy, and social support to enhance engagement in digital IC monitoring.
29 May 2026
Read appraisal →Journal of medical Internet research
Using the NASSS-Complexity Assessment Tool to Evaluate the Implementation of "Cadê O Kauê?": Chat-Story Intervention for Youth Participation in Mental Health Promotion in Brazil
BACKGROUND: The global public health challenge of young people's mental health is particularly evident in Brazil, which is considered the most anxious and the fifth most depressed country globally. In response to this, a digital intervention, that takes the form of a storytelling chatbot named "Cadê o Kauê?" (translation: "Where is Kauê?"), was coproduced by an interdisciplinary team of researchers, young people, and narrative designers as part of a project titled Engajadamente. The intervention was designed to strengthen Brazilian adolescents' skills in promoting their peers' mental health through direct support and collective action for mental health. The intervention was rolled out in select schools in Brazil and online via social media, but it encountered challenges during the initial implementation phase in school settings. OBJECTIVE: This project's primary objective is to evaluate the implementation of "Cadê o Kauê" in school settings in Brazil, assess current and potential complexities, and formulate recommendations to reduce or manage these complexities. The secondary objective of this project is to understand the current and potential sociotechnical change resulting from the implementation of the "Cadê o Kauê" in a real-world setting. METHODS: The Non-adoption, Abandonment, Scale-up, Spread, Sustainability (NASSS) framework was selected as the theoretical framework to identify different areas of complexity across domains and further thinking of ways to reduce or manage these complexities. Data were gathered by conducting semistructured online interviews with the Engajadamente team members based in the United Kingdom and Brazil. The NASSS-Complexity Assessment Tool was used as a sensemaking device to facilitate a collective understanding of the implementation, and narratives were mapped onto specific NASSS domains. Key themes were identified across overarching domains using thematic analysis to meet the objectives of this evaluation. RESULTS: The results mapped onto the NASSS domains generated a narrative about the initial implementation of "Cadê o Kauê?" in schools. Existing and potential complexities across technology, organization, and wider system domains were characterized by interdependencies, unintended consequences, and uncertainties. The implementation of "Cadê o Kauê?" highlighted challenges related to internet dependency, infrastructure limitations, and varying levels of organizational readiness, teacher preparedness, and school culture. Addressing these through offline solutions, additional teacher training, whole-school engagement strategies, and a value-informed complexity approach can enhance accessibility, alignment with school priorities, and adaptability within Brazil's dynamic sociopolitical context. CONCLUSIONS: Using the NASSS framework, the evaluation captured both current and emerging complexities in implementing "Cadê o Kauê?" in schools. By engaging with theory-informed approaches to technology implementation projects and deep-diving into discursive dynamic interactions between technology and its "context" of implementation, we have explored ways to manage these complexities and developed suitable recommendations to guide the Engadajamente team and support similar projects in the future.
29 May 2026
Read appraisal →PloS one
Adaptive and multi-scale feature fusion for Chinese news headline classification
The rapid growth of online news has led to an explosion of short Chinese headlines, which often suffer from sparse features, limited context, and high ambiguity-posing significant challenges for accurate classification. To address these issues, this paper proposes two tailored deep learning models: ERNIE-AAFF-SECNN for large-scale datasets, which enhances semantic representation via adaptive fusion of multi-layer ERNIE features and improves local feature extraction with SE-empowered CNN; and ERNIE-MSSE-DSCNN for small-scale datasets, which integrates multi-scale SE attention, depthwise separable convolutions, and adversarial training to boost robustness under data scarcity. A large number of experiments have shown that both of these models have achieved the most advanced performance. It is worth noting that the accuracy of ERNIE-AAFF-SECNN on the THUCNews and Toutiao datasets is 1.28% and 0.55% higher, respectively, than that of the lightweight SOTA model TinyBERT. The accuracy of ERNIE-MSSE-DSCNN on a 10% training dataset is 2.62% and 3.76% higher than that of the lightweight SOTA model TinyBERT, respectively. It demonstrates outstanding effectiveness under both standard and low-resource Settings. These results demonstrate that targeted architectural enhancements-such as adaptive feature fusion and multi-scale attention with adversarial training-can significantly improve the accuracy and robustness of short-text classification in practical Chinese news applications.
29 May 2026
Read appraisal →JMIR research protocols
An AI-Enabled Predictive Harm Response Management Algorithmic Tool to Reduce Adverse Events in Health Care Settings (PreHaRM): Protocol for a Three-Phase Model Development and Pilot-Testing Study
BACKGROUND: Current data analysis and coordination methods do not effectively support nurses and midwives in risk reduction, as retrospective reporting does not allow real-time insights and precludes proactive, preventive care. Analysis of administrative data within Australia's health care sector to predict risk may help address this shortcoming. Predictive analytics can transform these data into meaningful insights, identifying harm risk profiles that benefit the performance of Australian and international clinical programs. Importantly, these tools may support nurses and midwives in preventing adverse events and predicting high-risk situations. Researchers, in collaboration with local health network staff, will develop a proof-of-concept predictive risk algorithm. The "predictive harm response management algorithmic tool to reduce adverse events in healthcare settings" program (project DHCRC-0156) will provide real-time insights via an interactive dashboard, enabling nurses, midwives, and health administration to assess risks and optimize resources in health care settings. This protocol details the algorithm development activities for subproject 1a, predictive risk model development, which aims to develop and pilot-test a predictive harm algorithm across 2 South Australian local health networks. OBJECTIVE: This study aims to identify the clinical harm outcome of interest and relevant data sources per site and build a suitable data solution to model predictors of harm risk and identify actionable clinical, workforce, and environmental factors that affect the harm outcome of interest. METHODS: This study design includes three phases: (1) model generation, (2) model evaluation, and (3) prototype development. Data linkage by SA-NT DataLink can only proceed following approval from each of the following: the South Australia Department for Health and Wellbeing Human Research Ethics Committee, the University of South Australia Human Research Ethics Committee, and hospital governance committees. The clinical dataset will be split into a training set, a validation set, and a test set. Exploratory data analysis will be undertaken to ascertain features and classify outcomes from the raw dataset. Derived features will be computed, feature correlations will be estimated, and initial feature selection will be performed. Iterative model development will occur over 3 stages, and a dashboard to display these results will be developed. RESULTS: The study commenced on July 19, 2021, and will conclude on December 31, 2025. Finalized results are expected in December 2025. CONCLUSIONS: This research will conclude with the development of the algorithm for transferability to health care environments. Research activities will be detailed in publicly available reports and manuscripts prepared for peer-reviewed journals that will be drafted in accordance with existing and appropriate checklists. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/75474.
29 May 2026
Read appraisal →Journal of medical Internet research
Digital Literacy and Interpersonal Trust as Predictors of Willingness to Share Patient-Generated Health Data Among Korean Internet Users: Cross-Sectional Study Using Privacy Calculus and Communication Privacy Management Theories
BACKGROUND: The proliferation of wearable devices and advances in data analytics are accelerating the adoption of personalized digital health care, relying heavily on patient-generated health data (PGHD). However, the sensitive nature of this data creates significant privacy boundaries. While previous research has focused on rational cost-benefit trade-offs, there is a limited understanding of how social and cognitive factors-specifically interpersonal trust and digital literacy (DL)-shape the data-sharing decisions of the general public. OBJECTIVE: This study aims to identify the factors predicting individuals' willingness to share health data (WS) by integrating privacy calculus and communication privacy management theories. It specifically examines the comparative influence of DL, interpersonal trust, and moral motivation on data-sharing decisions. METHODS: We analyzed data from the 2023 Korea Panel Survey on the Digital Society (n=4518), a nationwide representative sample of internet users. Survey-weighted structural equation modeling with weighted least squares mean and variance adjusted estimation was used to examine the relationships among perceived risk (PR), perceived benefit (PB), DL, interpersonal trust, and WS. RESULTS: PR was the strongest negative predictor of WS (standardized coefficient [std β]=-0.189, P<.001), whereas PB was the strongest positive predictor (std β=0.076, P=.009), followed by moral motivation (std β=0.062, P=.03) and interpersonal trust. DL showed a significant negative direct association with willingness to share (std β = -0.060, P<.001). However, subdimension analysis revealed heterogeneous mechanisms: the "understanding" dimension was associated with lower PR and indirectly promoted sharing, whereas use and engagement were associated with higher PR. Age-stratified analyses suggested potential heterogeneity in these relationships, although the overall interaction was not statistically significant. CONCLUSIONS: PBs and risks were the strongest determinants of PGHD sharing, with benefits increasing and risks decreasing willingness to share. Beyond this risk-benefit balance, DL (particularly understanding) and interpersonal trust also played important roles, highlighting the need for trust-based and user-centric strategies to promote PGHD sharing and support the expansion of digital health care.
29 May 2026
Read appraisal →Journal of medical Internet research
Biocomputing: Beyond the Hype
Biocomputing is a nascent but rapidly developing field at the intersection of biology and computer science. In this News and Perspectives article, JMIR Correspondent Simon Spichak reports on its current and potential applications for health care research and beyond.
29 May 2026
Read appraisal →BMJ open
Voice-assisted artificial intelligence in cardiovascular disease management: a systematic review and meta-analysis protocol
INTRODUCTION: Cardiovascular disease (CVD) remains a leading cause of global morbidity and mortality, with self-management playing a pivotal role in improving outcomes. Voice-assisted artificial intelligence (AI) technologies such as virtual assistants and voice-controlled applications have emerged as innovative tools for healthcare delivery. While the technologies show promise in areas like primary prevention and chronic disease management, their effectiveness in supporting self-management for patients with CVD remains underexplored. This study aims to evaluate the impact of voice-assisted AI technologies on CVD self-management, specifically focusing on cardiovascular-related mortality, health-related quality of life (HRQoL) and adherence to lifestyle modifications. METHODS AND ANALYSIS: A systematic review and meta-analysis will be conducted following the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols. A comprehensive search will be performed across databases such as MEDLINE, Scopus, Embase and Cochrane Central Register of Controlled Trials (CENTRAL), from 2010 to 2025. The review will include randomised controlled trials (RCTs), non-RCTs and observational studies that evaluate voice-assisted AI interventions (eg, voice-controlled fitness apps, smart health assistants) aimed at CVD self-management. The primary outcome will be cardiovascular-related mortality. Secondary outcomes will include HRQoL, clinical outcomes (eg, high blood pressure), lipid profiles (eg, cholesterol and glucose levels) and lifestyle modifications (eg, dietary habits and levels of physical activity). Data management and analysis will be conducted using Comprehensive Meta-Analysis software V.2.0. The methodological quality of the included studies will be assessed using the Cochrane Risk of Bias tool for RCTs and the Newcastle-Ottawa Scale for observational studies. The meta-analysis will use random-effects models, with heterogeneity assessed using Q and I² statistics. Subgroup analyses and meta-regression will be conducted to explore potential sources of heterogeneity. ETHICS AND DISSEMINATION: No formal ethical assessment is required, as this study involves analysis of publicly available secondary data. Findings will be disseminated through publications in peer-reviewed scientific journals, conference presentations and media coverage to inform healthcare providers, policymakers and patients. PROSPERO REGISTRATION NUMBER: CRD42024568702.
29 May 2026
Read appraisal →European respiratory review : an official journal of the European Respiratory Society
Validating exacerbations of asthma in electronic health records: a systematic review
BACKGROUND: Previous studies have shown that the algorithms and code lists used to define asthma exacerbations vary across different sources of data, if reported at all. Defining and validating asthma exacerbations in electronic health records (EHR) would help to improve future research on asthma using EHR by leading to more consistent and comparable evidence. METHODS: We systematically reviewed the literature to evaluate studies that define exacerbations of asthma in EHR and report which algorithms have the highest validity. An adapted version of the QUADAS-2 designed for this review was used to assess risk of bias. RESULTS: Of the studies yielded by the search, only five met the inclusion criteria. Eligible studies used algorithms that contained codes from versions or modifications of either the 9th or 10th revisions of the International Statistical Classification of Diseases and Related Health (ICD-9 or ICD-10), and validity scores varied. Using the ICD-9 code 493 within algorithms to detect asthma exacerbations, sensitivity scores varied from 44.8% to 91.28% and specificity was >85%. Using the ICD-9 code 493.xx as the principal and secondary diagnosis in claims data, validity measures were all >85%. Using the ICD-10 code J45, scores for sensitivity, specificity and negative predictive value were also all >85%. CONCLUSIONS: Algorithms have been used to identify asthma exacerbations in EHR with varying degrees of validity. Algorithms including the ICD-9 code 493.xx or the ICD-10 code J45 to detect asthma exacerbations had high validity scores. However, there was a risk of bias in these studies and urgent work is needed using robust methods to validate definitions for future research using EHR.
29 May 2026
Read appraisal →BMJ open
Does a protocolised oxygenation target impact clinical outcomes in mechanically ventilated patients in the intensive care unit across the 10 academic and community hospitals in the UCHealth system? Protocol and statistical analysis plan for an EHR-embedded pragmatic stepped-wedge randomised controlled trial.
INTRODUCTION: We designed the Optimising Care in Critically Ill at the UCHealth by Liberalising the Target O₂ in Mechanically ventilated intensive care unit (ICU) Patients to determine the effectiveness of a multimodal educational and electronic health record order panel intervention in limiting occult hypoxaemia and hyperoxaemia in patients receiving invasive mechanical ventilation by targeting a prespecified oxygen saturation (SpO₂ 90%-96%) range. METHODS AND ANALYSIS: This trial is a pragmatic electronic health record-embedded, multisite, cluster-randomised, stepped-wedge implementation of a multimodal educational and electronic health record order panel intervention aimed at achieving a standardised intermediate target range of SpO₂ (90%-96%) or partial pressure of oxygen (PaO₂ 60-100 mm Hg) in mechanically ventilated adult patients admitted to the ICU across a 10 hospital health system that includes a large academic centre and smaller community hospitals.The primary endpoint is ventilator-free days to day 30, defined as the number of days alive and not receiving support through invasive mechanical ventilation following the first initiation of invasive mechanical ventilation in a participating ICU during the hospitalisation. Secondary outcomes include a variety of clinical endpoints: hospital-free days to day 30, all-cause mortality to day 90, need for supplemental oxygen at discharge and incidence of occult hypoxaemia and hypoxaemia. We will analyse primary and secondary endpoints using a mixed effects modelling framework, with specific distributions chosen depending on the type of outcome (eg, binary, count, ordinal, time-to-event). ETHICS AND DISSEMINATION: Research is performed under a waiver of informed consent for minimal-risk research, as approved by Colorado Multiple Institutional Review Board (24-0065). Results will be disseminated in peer-reviewed publications and at national and international conferences. TRIAL REGISTRATION NUMBER: NCT06501118.
28 May 2026
Read appraisal →JMIR medical education
Generative AI's Impact on the Mental Health of Medical Students: Scenario Analysis
BACKGROUND: Generative artificial intelligence (AI) is quickly changing medical education, even as medical students still face high levels of stress, anxiety, and burnout. These simultaneous trends-technological upheaval and ongoing mental health issues-bring up important questions about how future doctors will be trained and supported. Understanding how these factors might influence each other is crucial for developing resilient, future-ready medical education systems. OBJECTIVE: We carried out a foresight study using scenario analysis to examine potential futures at the crossroads of generative AI adoption and medical students' mental health. An initial environmental scan of the literature was conducted to pinpoint emerging trends and weak signals related to AI in medical education and well-being. These phenomena were categorized within a macro-meso-micro framework and analyzed through a multilevel sociotechnical change perspective. The study focused on 2 principal factors: the extent of generative AI integration into medical curricula and the availability of mental health support, as key drivers and critical uncertainties influencing future trajectories. METHODS: These dimensions resulted in 4 distinct scenarios: Analog Happiness (high support and low AI integration), Gen AI Paradise (high support and high integration), Disconnected Struggles (low support and low integration), and Gen AI Takeover (low support and high integration). Each scenario demonstrates how various institutional responses can impact students' digital readiness, psychological well-being, and professional growth. For each one, we identified the main systemic risks and suggested immediate institutional measures to address them. RESULTS: The findings suggest that technological innovation and mental health support must coevolve in medical education. Prioritizing one without the other risks producing either digitally unprepared or emotionally fragile physicians. Faculty readiness, ethical frameworks, and participatory curriculum design are critical to ensuring balanced integration. We formulated practical recommendations tailored to students, educators, and other stakeholders to guide balanced adaptation. CONCLUSIONS: Generative AI is more than just an additional tool in medical education; it is a systemic force that redefines how future physicians learn and operate. If technological change and student mental health are tackled separately, medical education risks creating graduates who are either unprepared for digital demands or mentally overwhelmed. This study highlights key systemic risks and suggests initial institutional steps to address them, providing a foresight-driven framework to assist educators and policymakers in responsible AI integration while safeguarding the well-being of future doctors.
28 May 2026
Read appraisal →Journal of medical Internet research
Artificial Intelligence in Patient-Centered Care and Macro-, Meso-, and Micro-Level Determinants of Rehumanization and Dehumanization: Qualitative Interview Study
BACKGROUND: Patient-centered care remains a foundational principle of modern health care. The digital transformation of health systems has accelerated the adoption of artificial intelligence (AI) across diagnostic, predictive, and communicative functions, with implications for efficiency and clinical workflows. At the same time, AI integration raises concerns regarding transparency, equity, accountability, and trust, positioning it as a potential driver of both rehumanizing and dehumanizing dynamics in health care practice. OBJECTIVE: This study examines how the adoption of AI in health care may influence patient-centered care, exploring its potential to promote rehumanization or contribute to dehumanization. The objective is to identify the factors that shape these outcomes at the macrolevel (policy and infrastructure), mesolevel (institutional practices), and microlevel (individual behaviors and interactions). METHODS: This study adopts an exploratory qualitative design informed by grounded theory principles, drawing on 20 semistructured interviews with health care leaders, clinicians, researchers, legal experts, and industry consultants who have substantial professional experience across European health care systems, with some participants also contributing experience from the US health care context. To enhance analytical rigor and transparency, the study applied the Gioia methodology, enabling inductive coding from first-order concepts to second-order themes and aggregate dimensions. This multistakeholder approach facilitated a nuanced examination of how AI integration is perceived and experienced across macro-, meso-, and microlevels of health care. RESULTS: The analysis identified key system-level factors shaping rehumanizing or dehumanizing outcomes of AI integration. At the macrolevel, 8 factors-including regulatory frameworks, policy priorities, and infrastructure-were identified as influencing whether efficiency pressures outweigh patient-centered values. At the mesolevel, 5 factors related to institutional strategies, workflows, and leadership shape how AI tools are embedded into care delivery. At the microlevel, 7 factors related to individual behaviors, trust, and doctor-patient interaction dynamics influence whether AI supports empathy and engagement or diminishes them. Rehumanizing potentials include reduced administrative burden, improved care pathways, clearer health communication, and enhanced decision-making, while risks include shorter consultations, reduced empathy, overreliance on automation, and erosion of professional identity. Without deliberate alignment with patient-centered principles, efficiency gains risk undermining the human dimensions of care. CONCLUSIONS: This study represents one of the first empirical examinations of how AI shapes health care practices through rehumanizing and dehumanizing dynamics. The findings demonstrate that outcomes depend not only on technical capabilities but also on regulatory frameworks, institutional strategies, and cultural adaptation. By systematically mapping influencing factors across macro-, meso-, and microlevels, the research provides actionable insights for decision-makers to ensure that efficiency gains remain aligned with patient-centered principles. Realizing AI's promise requires coordinated action to preserve empathy, trust, and interpersonal connection, ensuring that innovation strengthens rather than weakens the human dimensions of care.
28 May 2026
Read appraisal →JMIR medical informatics
Usability and Usefulness of Machine Learning-Based Clinical Decision Support Software in Primary Care: Survey of Users in a Prospective Observational Study
BACKGROUND: The successful implementation of decision support systems promises to enhance high-quality care. However, the successful implementation of a clinical decision support system (CDSS) depends on user acceptance and adoption. A machine learning (ML)-based CDSS to assist primary care professionals treating urinary tract infections (UTIs) was implemented, and usability and usefulness were assessed through a questionnaire. OBJECTIVE: This study aimed to assess the system's usability by examining users' experiences with the software. A secondary goal was to assess users' attitudes toward evidence-based practice and innovation in health care. METHODS: In collaboration with the Netherlands Institute for Health Services Research (NIVEL) and Leiden University Medical Center (LUMC), Pacmed Ltd developed the CDSS. The cohort was mostly recruited at the care group level; practices within participating care groups were required to participate. Health insurers partly funded the research. Practitioners participated in the implementation study for 4 months. A survey based on the Unified Theory of Acceptance and Use of Technology (UTAUT) was sent to 263 general practitioners and assistants shortly after the implementation period. Furthermore, usage data were analyzed. RESULTS: Of the 34 participating practices that used the software, 30 (88%) submitted at least one survey response, with a mean of 2.23 responses per practice (SD 1.43). The CDSS was used throughout the pilot period, and 31 practices continued using the tool, with 9% dropping out during the first 8 weeks. Sixty-seven percent of respondents trusted the tool's output, and 73% found it understandable how the algorithm came to predictions. Sixty-five percent of respondents indicated that the information provided was useful in addition to the available guidelines, and 52% agreed that it supported their decision-making. However, many respondents were uncertain whether the tool improved patient care (46%) or patient outcomes (66%). Forty-eight percent of respondents found the software easy to integrate into their clinical workflow. CONCLUSIONS: The CDSS was perceived as trustworthy and easy to use. However, users were unable to determine whether the CDSS improved patient outcomes. In addition, the CDSS development could have benefited from including assistants as well as general practitioners more in the design phase of the software. Because assistants play an important role in UTI care, designing the software to better fit existing workflows may reduce the perceived time investment associated with using the tool. Finally, respondents reported strong motivation to contribute to further research in this field and indicated willingness to embrace change in health care delivery, which may also reflect selection bias in our sample.
28 May 2026
Read appraisal →European radiology experimental
Who labels best? Radiologists, rules, or large language models for CT reports on pulmonary embolism.
OBJECTIVE: To compare open-weight and proprietary large language models (LLMs), a rule-based extractor (RBE) and radiologists for labelling pulmonary embolism CT reports, and to test whether a hybrid RBE-LLM workflow improves labelling performance. MATERIALS AND METHODS: This single-centre retrospective study included structured CT reports from October 2021 to March 2025. Three labelling pipelines were evaluated: an RBE; a model-agnostic LLM extractor (18 open-weight, four GPT-4 variants); and a hybrid pipeline routing only RBE failures to an LLM. Ground truth was defined at the report-text level by deterministic schema matching for initially RBE-valid fields and blinded adjudication of RBE-invalid fields by two attending radiologists. Eight radiologists provided a human baseline. Outcomes included F1 scores, accuracy, LLM-based salvage of RBE failures, and labelling time. RESULTS: In total, 2,923 reports from 2,923 patients (mean age 66 ± 17 years; 1,465 women) were included. Falcon3-10b and GPT-4.1-mini achieved similar item-level performance (F1 0.98 [95% CI, 0.97-0.98] for both; p = 0.70) and both exceeded the RBE (F1 0.81 [95% CI, 0.80-0.82]; p < 0.001). Salvage of RBE failures was comparable between open-weight and proprietary models (88.1% vs 91.9%; p = 0.12). The hybrid RBE-LLM workflow achieved 99.8% accuracy and F1 0.99 (0.98-0.99), exceeding both the RBE and pooled radiologists (F1 0.92 [95% CI, 0.90-0.93]; all p < 0.001). CONCLUSION: Schema-constrained open-weight and proprietary LLMs exceeded rule-based extraction and, at the upper end of performance, matched a pooled radiologist label-transfer baseline. A rules-first, targeted LLM workflow enabled near-perfect extraction from finalised structured pulmonary embolism CT reports. RELEVANCE STATEMENT: A rules-first LLM workflow can automate high-fidelity extraction of structured CT findings from finalised radiology reports, enabling scalable, auditable, and more consistent cohort curation for clinical research, registries, and quality improvement. KEY POINTS: A hybrid rules-first workflow combining a rule-based extractor (RBE) with targeted large language model (LLM) salvage achieved the highest overall performance for labelling of pulmonary embolism CT reports (F1, 0.99; accuracy, 99.8%). The top standalone open-weight and proprietary LLMs (Falcon3-10b and GPT-4.1-mini) both exceeded the RBE and, at the upper end of performance, matched a pooled radiologist label-transfer baseline. The hybrid workflow reduced cohort-curation time from 32.2 h for radiologists to 1.0 h while reducing LLM calls by 85.6%, because the LLM was only triggered for rule-failed fields.
28 May 2026
Read appraisal →JMIR formative research
Differences in Safety Risks Across Languages in Health-Relevant Queries: Vulnerability Analysis of Large Language Model Responses
BACKGROUND: Large language models (LLMs) such as ChatGPT are increasingly used to support health-related queries and decision-making. However, these models can be "jailbroken" through adversarial prompts that bypass safety filters and elicit harmful or medically inappropriate responses. In health care contexts, such vulnerabilities pose serious risks. Understanding how jailbreak susceptibility varies across languages is essential for developing robust safeguards and promoting equitable access to safe health information. This paper may contain examples that may be deemed harmful in terms of violence, self-harm, and drug abuse. OBJECTIVE: This study aims to systematically compare and contrast the vulnerability of a health LLM for jailbreaking across 3 languages: English, Spanish, and Hindi (transliterated using the Latin alphabet), based on emoji and permutation cipher attacks. METHODS: We analyzed 1000 input prompts per language, drawn from the BeaverTails dataset, across 3 harm categories: self-harm, violence, and drug abuse. Each prompt was modified using emoji and permutation cipher techniques, resulting in 6000 input-output pairs. Model responses were evaluated by human coders to determine the success rate of jailbreak attempts across languages and cipher types. RESULTS: Hindi prompts showed the highest vulnerability, with 787 successful jailbreaks using emoji ciphers and 873 using permutation ciphers. Spanish and English followed, with lower success rates across both cipher types. Differences in jailbreak success across languages and cipher strategies were statistically significant. Additionally, attacks targeting violence-related prompts were more successful overall than those targeting drug-related or self-harm content, indicating variation in vulnerability by harm type. CONCLUSIONS: The findings of this formative study reveal that LLM safety performance varies substantially across languages and harm categories, raising concerns about equitable protection in multilingual health communication. Disparities in access to harmful content may contribute to downstream health risks. Strengthening multilingual content moderation and developing language-aware safety mechanisms are critical steps toward creating safer and more inclusive health AI systems.
28 May 2026
Read appraisal →JMIR mental health
When AI Colludes: Clinical Reliability of Training and Preference Data as a Trustworthy-AI Criterion
Research on artificial intelligence (AI) and mental health has focused largely on harms at deployment, including chatbot safety, sycophancy, and AI-associated delusions. Less attention has been paid to a prior question: whether the human-generated text and preference judgments that shape large language models are themselves clinically reliable, particularly when self-report may be distorted. This Viewpoint aims to develop the clinical psychiatric construct of collusion-the uncritical acceptance of an unreliable account-as an analytic lens for AI training and deployment, and to argue that the clinical reliability of training and preference data should be treated as an explicit trustworthy-AI criterion in mental-health-relevant systems. A conceptual synthesis of psychiatry, clinical psychology, and AI safety literature was undertaken. The analysis distinguishes three pipeline layers: pretraining corpora, preference data and posttraining methods, and deployment-time interaction. It maps the clinical construct of collusion against adjacent technical concepts, including sycophancy, reward overoptimization, grounding, refusal training, red-teaming, and live monitoring. The synthesis suggests that collusion-like dynamics are least applicable at the pretraining layer and most applicable at the preference-data and deployment layers, where unassessed user or labeler input can be reinforced without corroboration. Existing mitigations, including data curation, Constitutional AI, reward-model evaluation, grounded generation, refusal training, red-teaming, and postdeployment monitoring, address parts of this problem. However, these approaches are not yet organized around a clinically informed account of when self-report is unreliable. The central novelty is therefore not a generic claim about bias, but the proposal that clinical self-report reliability should be assessed as a distinct data-quality and governance dimension. Trustworthy-AI frameworks for mental-health-relevant applications should incorporate clinical expertise in self-report reliability into preference-data design, red-teaming, and postmarket surveillance. Adding the clinical reliability of training and preference data as an explicit criterion could complement existing technical safeguards while leaving empirical evaluation of clinician involvement as an open research agenda.
28 May 2026
Read appraisal →JMIR research protocols
Coconstructing CHAMP, an Artificial Intelligence Chatbot for Pediatric Infectious Symptoms Management: Protocol for a Multiphase Participatory Study
BACKGROUND: Acute infectious symptoms are a leading cause of pediatric emergency department visits in Canada, many of which are low acuity and could be safely managed at home. Artificial intelligence (AI) chatbots offer a promising avenue for delivering accessible, evidence-based guidance to support families in managing these symptoms. OBJECTIVE: This study aims to adapt and coconstruct CHAMP (CHatbot to Assist the Management of Pediatric patients), an AI chatbot to support patients and families with acute pediatric infectious symptoms. CHAMP aims to deliver timely, tailored, and validated health information to support safe at-home self-management and informed care-seeking. METHODS: This multiphase, mixed methods participatory study will be conducted at the Montreal Children's Hospital in Montreal, Quebec, Canada. A coconstruction committee comprised of youth, parents, caregivers, and partners will be engaged as coresearchers. Eligible participants will include (1) youth aged 14-17 years and (2) parents and caregivers of children aged 0-17 years. The study comprises 5 phases. Phase 1 involves a qualitative needs assessment using focus groups with 20 participants to explore their informational needs, preferences, and concerns regarding pediatric infections and the use of AI chatbots. Phase 2 focuses on coconstructing and validating CHAMP's knowledge database through 3-5 workshops. Coresearchers will review pediatric clinical guidelines, map care questions and decision-making processes, and shape CHAMP's conversational framework. Phase 3 consists of iterative prototyping and testing through 3-5 workshops. Coresearchers will engage in prototyping and scenario testing, alongside preliminary usability and acceptability assessments. Phase 4 examines equity and accessibility through focus groups with 20 participants at risk of digital exclusion, as well as multilingual evaluation of an automated large language model-based translation layer. Phase 5 uses collaborative ethnography to explore the process of participatory coconstruction and its impact on CHAMP's design. RESULTS: Funding was secured in 2024, and Research Ethics Board approval was obtained in December 2024. As of December 2025, the coconstruction committee is being assembled, and Phase 1 recruitment is underway. CONCLUSIONS: This study will produce a functioning CHAMP prototype grounded in participatory, equitable, and responsible pediatric AI development. Findings will inform usability testing and an implementation-effectiveness evaluation, contributing to best practices for pediatric-centered AI health tools. By providing timely, tailored, and validated health information on acute infections, CHAMP may support safe at-home self-management, reduce preventable emergency department visits, ensure at-risk children are directed to appropriate care, and improve patient and family health care experiences.
28 May 2026
Read appraisal →Journal of medical Internet research
Blockchain-Enabled Self-Sovereign Identity Applications in Health Care: Scoping Review
BACKGROUND: Self-sovereign identity (SSI) provides a decentralized approach to digital identity management, enabling individuals to control their personal data without reliance on centralized authorities. Blockchain technology offers a tamper-resistant and distributed infrastructure that can support secure and verifiable identity systems. In health care, where identity fragmentation, privacy risks, and interoperability challenges persist, blockchain-enabled SSI (BC-SSI) has been proposed as a potential solution. However, existing research remains heterogeneous, with varying levels of technical maturity and limited evidence of real-world deployment. OBJECTIVE: This study conducts a scoping review to systematically map BC-SSI applications in health care and to analyze their application domains, development stages, study aims, targeted challenges, and technological infrastructures. In addition, this study aims to identify structural gaps in current research and assess the readiness of BC-SSI systems for clinical deployment. METHODS: This review followed the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) methodology. A comprehensive literature search conducted between September 2024 and August 2025 identified 37 peer-reviewed studies that met predefined inclusion criteria. Data were extracted and synthesized using descriptive and thematic analyses across application areas, system maturity, technological components, and reported challenges. RESULTS: The findings indicate that BC-SSI research in health care remains at an early stage of maturity, with most studies proposing conceptual models or prototype implementations and limited real-world validation. Applications predominantly focus on identity verification, credential management, and privacy-preserving data exchange across domains such as electronic health records, mobile health, and access control systems. Commonly used technologies include decentralized identifiers, verifiable credentials, smart contracts, and privacy-enhancing mechanisms such as zero-knowledge proofs and selective disclosure. Despite rapid technical development, persistent challenges include interoperability limitations, governance gaps, usability concerns, and insufficient integration with health care infrastructures. Notably, a structural gap was identified between technological capability and system-level readiness for clinical deployment. CONCLUSIONS: BC-SSI technologies demonstrate potential for enabling secure, interoperable, and patient-centric identity management in health care. However, current research is predominantly technology-driven and lacks sufficient system-level validation. This study highlights the need for integrated architectural approaches, governance frameworks, and real-world evaluation to bridge the gap between conceptual innovation and clinical implementation. Advancing BC-SSI toward health care adoption will require coordinated progress across technical, organizational, and regulatory dimensions.
28 May 2026
Read appraisal →PloS one
A hybrid machine learning and explainable AI framework for optimizing risk-based authentication
As online platforms continue to grow, the need for strong authentication mechanisms becomes increasingly important to protect sensitive information and networks. Risk-Based Authentication (RBA) is an adaptive approach that dynamically adjusts authentication decisions based on user behavior and contextual information, thereby improving both security and user experience. This study proposes a hybrid RBA framework that integrates machine learning ensemble techniques, fuzzy logic, clustering, and optimization to enhance account takeover detection and dynamic risk assessment. The ensemble classifier, combining Gradient Boosting, SVM, and XGBoost, predicts the probability of account compromise based on login behavior, device attributes, and network information. K-Means clustering is used to generate initial risk thresholds (low, medium, and high), which are further refined using a fuzzy logic system to map probabilities to risk levels. The L-BFGS-B optimization algorithm is employed to fine-tune fuzzy membership boundaries and improve threshold consistency. Experimental results demonstrate strong performance, achieving 97.77% accuracy, 99.41% precision, 98.04% recall, 98.72% F1-score, and an EER of 0.0303. On large-scale datasets ranging from 2M to 30M records, the proposed framework demonstrates consistent improvement in authentication decisions. For the 2M dataset, Allow Login actions increase from 349,432-349,923, while Deny Login actions decrease from 1,462-1,228, along with a slight reduction in additional authentication prompts. Furthermore, the use of Explainable AI techniques, particularly SHAP, enhances the transparency and interpretability of the model, supporting more informed decision-making. Overall, the proposed framework is accurate, adaptive, and suitable for real-world risk-based authentication applications.
27 May 2026
Read appraisal →Journal of radiation research
Feasibility of retrieval-augmented generation for large language models with Japanese input in radiotherapy
Large language models (LLMs) have recently gained attention for their potential. However, concerns remain regarding their reliability due to limitations such as hallucinations and insufficient domain-specific knowledge. Retrieval-augmented generation (RAG) has emerged as a promising approach, enabling LLMs to reference external knowledge sources and generate accurate outputs. We aimed to clarify the potential of RAG-enhanced LLMs with Japanese input in the field of radiotherapy. This was assessed by evaluating performance on three certification examinations in Japan: the Japanese Medical Physicist Examination, the Japanese Board Examination for Radiologists, and the Japanese Board Examination for Radiation Oncologists. In this study, we constructed a RAG system named Rad-Hub, consisting of a Japanese Radiotherapy Knowledge Database (JRKD) and a retrieval framework built on Microsoft Azure. The JRKD was populated with 32 Japanese radiotherapy textbooks and clinical guidelines. We assessed its utility by inputting all multiple-choice questions from the three examinations into ChatGPT-4o, both with and without Rad-Hub, and recording the answers. They were then compared with reference answers determined by experienced medical physicists and radiation oncologists. Rad-Hub improved accuracy across all examinations. Accuracy increased from 77.0% ± 2.6% to 84.6% ± 1.5% in the Medical Physicist examination, from 74.9% ± 2.0% to 82.1% ± 1.1% in the Radiologist examination, and from 55.6% ± 4.4% to 71.7% ± 4.5% in the Radiation Oncologist examination. Performance gains ranged from 7.2% to 16.1%. These findings highlight the potential of RAG-enhanced LLMs, particularly ChatGPT-4o with Rad-Hub, for integration into radiotherapy applications, such as educational and clinical decision assistance.
27 May 2026
Read appraisal →Journal of medical Internet research
Sequencing AI Automation and Data Interoperability in Oncology Using a Scenario-Planning Framework Coupled With Discrete-Event Simulation: Proof-of-Concept Study
BACKGROUND: As oncology workflows integrate increasingly autonomous artificial intelligence (AI) agents, health systems face uncertainty regarding operational impacts. Traditional linear forecasting methods fail to capture second-order effects such as governance saturation, induced demand, and bottleneck migration. To navigate this complexity, the emerging field of medical futures studies requires methodologies that bridge qualitative strategic foresight with quantitative operational modeling. These system-level dynamics directly influence timely diagnosis, treatment delays, and overall health system resilience. OBJECTIVE: This study aimed to develop a proof-of-concept framework coupling qualitative scenario planning with computational discrete-event simulation to stress-test oncology AI adoption strategies. METHODS: We defined a strategic state space using 2 orthogonal axes, AI automation intensity and data interoperability, resulting in 4 distinct futures scenarios. We translated these qualitative narratives into a quantitative discrete-event simulation model of a 3-year operational horizon. The model quantified system performance (referral-to-treatment interval [RTTI] and throughput), volatility, and resource constraints across different adoption trajectories. RESULTS: The scenario-planning phase yielded 4 operational archetypes (analog oncology, automation islands, interconnected clinicians, and AI-orchestrated care) with distinct constraints, risks, and failure modes. In the simulation, the fully integrated scenario maximized capacity (1244, SD 21.4 patients per year) and halved the mean RTTI to 14.9 (SD 0.3) days, a magnitude comparable to major pathway redesign interventions. Isolated automation without data infrastructure led to reduced system performance, increasing RTTI by 26% (37.1, SD 1.3 days) and reducing throughput to 647 (SD 10.1) patients per year due to administrative governance saturation. The model illustrated a structural bottleneck migration: successful upstream AI adoption shifted binding constraints from diagnostic scanners to downstream chemotherapy infusion units, whereas missing data interoperability resulted in governance constraints. Pathway optimization analysis indicated that a coordinated strategy prioritizing early improvements in data interoperability reduced transition volatility compared to an automation-first approach. CONCLUSIONS: Integrating qualitative scenario planning with quantitative simulations enabled a systematic evaluation of oncology AI adoption strategies. As a proof of concept, it offers a replicable framework for health leaders to model future scenarios of digital transformation in times of high uncertainty. Subsequent work should expand this methodology to incorporate financial and health equity dimensions, establishing simulation-based scenario planning as an important tool in medical futures studies.
26 May 2026
Read appraisal →Clinical chemistry and laboratory medicine
Automation in value-based laboratory medicine: driving precision, capacity, and better outcomes
Laboratory medicine lies at the core of modern healthcare, enabling timely diagnosis, effective patient monitoring, and increasingly personalized therapeutic strategies. Over the past decades, automation has profoundly reshaped the role of clinical laboratories, substantially enhancing their contribution to clinical outcomes, operational efficiency, and the overall sustainability of healthcare systems. More recently, laboratory automation has emerged as a cornerstone of value-based laboratory medicine, representing not merely a technological upgrade but a strategic transformation of laboratory practice aimed at delivering measurable value to patients and healthcare stakeholders. Although automation has long been established in clinical chemistry and immunoassays, its scope is now expanding to molecular diagnostics and mass spectrometry - two disciplines that are central to precision medicine. Looking ahead, the convergence of automation, digitalization, and artificial intelligence is driving the emergence of hyperautomation in laboratory medicine. Within this paradigm, laboratories evolve from isolated testing units into integrated diagnostic hubs, in which results from multiple laboratory disciplines are harmonized and contextualized to effectively support clinical decision-making.
26 May 2026
Read appraisal →JMIR formative research
An In-Hospital Mortality Risk Model for Patients Undergoing Coronary Artery Bypass Grafting Based on Machine Learning: Cohort Study
BACKGROUND: Ischemic heart disease remains the leading cause of death worldwide. Coronary artery bypass grafting (CABG) remains the primary surgical treatment for ischemic heart disease. There is currently a lack of highly accurate and widely applicable models for assessing the risk of postoperative mortality following CABG. OBJECTIVE: This study aimed to develop and validate an in-hospital mortality risk prediction system for patients undergoing coronary artery bypass grafting (CABG) by using machine learning algorithms and to compare its performance with the European System for Cardiac Operative Risk Evaluation II (EuroSCORE II) and Sino System for Coronary Operative Risk Evaluation (SinoSCORE). METHODS: Between January 2017 and December 2020, 21,443 patients undergoing CABG in the Chinese Cardiac Surgery Registry were included. Patients were randomly divided into training (n=17,753) and test (n=3690) cohorts. We addressed class imbalance using the synthetic minority oversampling technique (SMOTE) and optimized hyperparameters via grid search. Fifteen machine learning algorithms were developed to predict in-hospital mortality. Performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration metrics (Brier score), and decision curve analysis, and was compared against EuroSCORE II and SinoSCORE. RESULTS: A total of 21,443 patients were included. Overall, in-hospital mortality was 2.1% (n=450). The Extreme Gradient Boosting (XGBoost) model achieved the best performance with an AUC of 0.850 in the training cohort and 0.782 in the independent test cohort (this cohort was independent and not involved in model construction). While EuroSCORE II showed an AUC of 0.722 and SinoSCORE showed an AUC of 0.726 in the test cohort, the XGBoost model demonstrated superior discrimination and calibration (P<.05). CONCLUSIONS: Our study developed and validated a machine learning-based risk prediction model for in-hospital mortality after CABG by using a large-scale Chinese multicenter registry. Among the algorithms tested, the XGBoost model demonstrated superior discrimination and calibration compared with the traditional EuroSCORE II and SinoSCORE, suggesting that locally calibrated models may better capture the risk profile of Chinese patients. The derived 7-variable web calculator may serve as an exploratory auxiliary tool to provide a preliminary reference for bedside risk stratification, though its direct impact on surgical decision-making requires further prospective validation. Future research should focus on independent test cohorts across diverse hospital tiers to ensure broad generalizability.
26 May 2026
Read appraisal →Journal of medical Internet research
Exploring Technological Solutions for Interoperability Between Patient Electronic Medical Records and Clinical Registries: Scoping Review
BACKGROUND: The use of electronic medical records (EMRs) and clinical registries has transformed health care delivery by improving data management, care coordination, and research capacity. However, the full potential of these technologies can only be realized through effective interoperability, thereby reducing the burden of manual data entry and enhancing the use of real-world clinical data. OBJECTIVE: This review examines technologies that enable automated data extraction and transfer, which promote interoperability between EMRs and clinical registries. METHODS: A search of PubMed, CINAHL, Embase, and Web of Science, including studies published between January 2013 and April 2025, was registered with Open Science Framework a priori and involved three key concepts: (1) "registry," (2) "electronic medical records," and (3) "interoperability." A 2-phase screen identified studies evaluating technologies that facilitate automated data extraction or interoperability. Automation was defined as fully automated, where data are extracted and transferred without human intervention, or semiautomated, where extraction or transfer is predominantly automated but may include manual validation. Only technologies supporting ongoing database integration were eligible for inclusion. Screening, data extraction, and synthesis were conducted by multiple independent reviewers. Technology experts provided extensive input and guidance throughout to ensure the accuracy and relevance of the extracted information. RESULTS: Overall, 36 studies met the inclusion criteria, representing 12 countries across 5 continents and addressing a wide range of acute and chronic health conditions. Epic was the most frequently reported EMR system, while the most common registry platforms were REDCap (Research Electronic Data Capture; Vanderbilt University), structured query language (SQL) server database, and EMR-embedded solutions. Most approaches centered around extracting data from structured formats (n=18), or a combination of both structured and unstructured formats (n=10), emphasizing the central role of structured EMR data in current automated extraction approaches. CONCLUSIONS: This review advances understanding of interoperability between EMRs and clinical registries by uniquely examining automated and sustainable solutions for data exchange, extending beyond prior work that has largely focused on technologies designed for isolated systems or study-specific data extraction. A novel contribution of this review is the synthesis of context-specific considerations derived from reported implementations, providing a comprehensive overview of how technology selection and implementation are shaped by the context in which they are deployed. While these advancements have reduced reliance on inefficient, error-prone, and resource-intensive manual processes, ongoing challenges in data standardization, seamless integration, and long-term sustainability are compounded by poor and inconsistent reporting across studies. Future efforts should follow comprehensive reporting guidelines, adhere to robust governance principles, and incorporate implementation science frameworks, to not only enable meaningful comparison and synthesis in future research, but also to ensure that technologies can be effectively, feasibly, and sustainably integrated within health care contexts, while upholding the ethical and equitable use of health care data.
26 May 2026
Read appraisal →Journal of medical systems
A Standardized Framework for Evaluating State Opioid Surveillance Dashboards: A Health Informatics Assessment
BACKGROUND: State public health agencies across the U.S. depend on opioid surveillance dashboards to monitor overdose trends and guide interventions. Despite their central role in decision-making, no standardized framework exists to evaluate how well these dashboards actually function. OBJECTIVE: We developed a health informatics framework to systematically assess state opioid surveillance dashboards and applied it across all 50 U.S. states. METHODS: Our framework evaluates dashboards across nine domains (including temporal analysis, geographic granularity, data disaggregation, visualization features, narrative support, simplification techniques, and interactive functionality) using 28 specific sub-variables. Sub-variables were selected based on public health informatics literature on effective dashboard design, including features identified in COVID-19 dashboard evaluations and usability frameworks. We scored each feature as either present or absent, creating composite scores with a theoretical maximum of 28 points. All 50 U.S. state-level dashboards were reviewed twice by the study team, with an external consistency check on a subset of 10 dashboards. We calculated the correlation coefficient between dashboard scores and state population estimates. RESULTS: Dashboard scores varied widely, ranging from 12 to 22 points (mean:19.18, standard deviation:2.20). Three states (Delaware, Indiana, and New York) achieved the highest score of 22 by offering comprehensive geographic detail, extensive data breakdowns, and full interactive capabilities. Interactive filtering was present in 92% of dashboards, but only 80% enabled switching between visualization modes. While nearly all dashboards showed time trends, only 6% provided zip code-level detail, and over a third lacked race and ethnicity data. Single external reviewer demonstrated strong correlation (r = 0.84, p = 0.002). Dashboard scores were not correlated with state population size (r = 0.187, p = 0.193). CONCLUSIONS: State opioid surveillance dashboards varied substantially in observable capabilities. This framework provides a standardized method to compare dashboard features and identify gaps that may inform future refinement.
25 May 2026
Read appraisal →JMIR formative research
Prediction of 30-Day All-Cause Hospital Readmissions Using Limited Structured Electronic Health Record Data: Retrospective Comparative Study
BACKGROUND: Unplanned hospital readmissions represent a critical operational and financial challenge for health care systems in the United States, with 3.8 million 30-day all-cause readmissions in 2018 at an average cost of US $15,200 each, totaling US $58 billion in costs. Many published prediction models rely on comprehensive information (eg, full billing abstractions, discharge summaries, laboratory tests, and vitals) that becomes available only late in the encounter, limiting usefulness for real-time, in-hospital intervention. This creates a timeliness-accuracy trade-off: models that are most accurate retrospectively may arrive too late to act upon. OBJECTIVE: This study tests whether a clinically meaningful predictive signal for 30-day all-cause readmission is present within a limited set of structured clinical codes recorded during the patient's hospital stay. This approach evaluates whether predictive signals are retained when using a restricted set of structured clinical codes. METHODS: We conducted a retrospective comparative modeling study using a large, deidentified electronic health record dataset of 50,000 inpatient encounters from the 2019 New York State Emergency Department Database. Two feature sets were constructed: (1) a limited set consisting of the first 5 ICD-10 (International Classification of Diseases, 10th Revision) diagnosis codes, the first 5 Current Procedural Terminology (CPT) codes, and Charlson Comorbidity Index (CCI; 11 input features); and (2) a rich set using all available ICD-10 and CPT codes plus CCI (up to 135 input features). We trained 4 models: random forest, CatBoost, multilayer perceptron, and DistilBERT (a distilled Bidirectional Encoder Representations from Transformers [BERT] model; structured codes mapped to text and tokenized with DistilBERT-base-uncased). Evaluation used an untouched hold-out set of 10,000 encounters, preserving the natural 21.1% readmission prevalence. Primary metrics were area under the receiver operating characteristic curve (AUROC), F1-score, and accuracy. To address class imbalance, the training split only was balanced via undersampling of the majority class and bootstrap oversampling of the minority class; validation/test distributions were left unchanged. RESULTS: Models trained on the limited feature set achieved AUROC values ranging from 0.5369 to 0.5596 and F1-scores from 0.2555 to 0.3434. Across 3 of 4 architectures, models trained on the limited feature set matched or exceeded the discrimination of their rich counterparts. The best model (random forest, limited) achieved an area under the curve AUROC 0.5596 (95% CI 0.5440-0.5739) compared to the best performing rich model (DistilBERT) at 0.5703 (95% CI 0.5565-0.5842), an absolute difference of 0.0107. The highest F1-score (0.3434) was achieved by DistilBERT on the limited feature set. Differences across architectures were small in absolute terms, with threshold-dependent metrics (eg, F1-score) being comparable. CONCLUSIONS: The findings suggest that models using a limited set of structured clinical codes can achieve performance comparable to those using more comprehensive coding information.
25 May 2026
Read appraisal →International journal of circumpolar health
Artificial intelligence/machine learning in Alaska Native healthcare systems: symposium perspectives.
Artificial intelligence and machine learning (AI/ML) utilisation in Alaska Native and American Indian (AN/AI) healthcare presents an opportunity to address persistent health disparities and improve the well-being of Alaska Native communities. The 2025 Artificial Intelligence in Alaska Native Health Care Systems symposium occurred in Anchorage, Alaska, with 31 individuals participating in facilitated conversations between various stakeholders and specialists at the Alaska Native Tribal Health Consortium, Southcentral Foundation, Maniilaq Association and Stanford University. The overarching goal of the symposium was to bring together partners within the Alaska Tribal Health System (ATHS) to promote: (1) AN/AI peoples' guidance in support of Tribal sovereignty for AI/ML implementation within the ATHS, (2) expanded technical capacity and (3) centralised guidance and expertise to lead future AI/ML research and implementation with community-defined benefits. During the event, small group discussions focused on the following topics: (1) common AI/ML applications, (2) concerns surrounding AI/ML, (3) current or potential policies guiding AI/ML use and (4) how to build trust in AI/ML models. Themes discussed were data democratisation, AI/ML development and use needs to occur under the principles of data sovereignty, and AI/ML quality improvement and research are needed and distinct. The symposium was an important step in considering how to use AI/ML to optimise AN/AI health and well-being through collaborative partnerships and services.
25 May 2026
Read appraisal →Behavior research methods
Automating data extraction in meta-research: A multi-model benchmark in network psychometrics papers
Manual data extraction in meta-research is often tedious, time-consuming, and error-prone. In this paper, we investigate whether the current generation of large language models (LLMs) can be used to extract accurate information from scientific papers. Across the meta-research literature, these tasks usually range from extracting verbatim information (e.g., the number of participants in a study, effect sizes, or whether a study is preregistered) to making subjective inferences. Using a publicly available dataset containing a wide range of metascientific variables from 43 network psychometrics papers, we tested five LLMs (Claude 4.6 Opus, Claude 4.5 Sonnet, Claude 4.5 Haiku, GPT-5.2, and GPT-5 mini). We used an automated API-based pipeline to extract variables from the documents. This approach allows batch processing of research papers. As such, it represents a more efficient and scalable way to extract metascientific data than the default chat interface. The extraction accuracy ranged from 79.6% to 91.3% across the models. The extraction performance was generally higher for more explicit, verbatim information and worse for variables that required more complicated inference. Furthermore, most models were able to convey uncertainty in more contentious cases. We provide a comparison of the accuracy and cost-effectiveness of the individual models and discuss the characteristics of variables that are and are not suitable for automatic coding. Furthermore, we describe some of the common pitfalls and best practices of automated LLM data extraction. The proposed procedure can substantially reduce the time and costs associated with conducting meta-research.
25 May 2026
Read appraisal →Nicotine & tobacco research : official journal of the Society for Research on Nicotine and Tobacco
Training a Smoking Status Probabilistic Model Using Cotinine Levels in a Large Claims Database
INTRODUCTION: Smoking status is an important confounder for many epidemiologic studies, yet it is not well documented in common sources of real-world data, including administrative claims. Probabilistic models can be used to create a proxy for smoking status, yet most published models have been trained using self-reported data. The objective of this study was to train a smoking status probabilistic model using cotinine values available in a large claims database. METHODS: Beneficiaries were included if they had at least one cotinine measurement and were categorized as a "current smoker" if their serum or plasma cotinine value was ≥5 ng/mL or urine cotinine value was ≥30 ng/mL. Predictors were collected across one year prior to the cotinine assessment date. The model was fit using logistic regression with stepwise forward selection. Model performance was assessed using discrimination and calibration. RESULTS: The final model yielded an area under the receiver operating characteristic curve of 0.77 (95%CI:0.75-0.78) and was well calibrated across most prediction deciles. The strongest predictors included diagnosis codes for smoking and drug abuse, and number of medications. The model was found to be highly specific, yet not sensitive at probability cutoffs ≥0.2. CONCLUSIONS: A smoking status model was developed and internally validated for application in claims data, using available cotinine values to define smoking status and found to have acceptable discrimination and calibration. The model is based on 26 predictors, fewer than other similar published smoking status models. External validation of the model should be a next step toward utilizing the model for epidemiological research. IMPLICATIONS: This study tests the utility of cotinine values to validate a smoking status probabilistic model, which has not been done in the literature to date. The results were robust to various cotinine levels used to define smoking status, per current guidance. The final model uses only 26 factors to predict smoking status, simplifying the application of the model in other claims databases.
25 May 2026
Read appraisal →Calcified tissue international
Performance of Large Language Models in Answering Osteoporosis-Related Frequently Asked Questions: A Systematic Comparative Evaluation Based on International Associations
To systematically evaluate and compare the performance of large language models (LLMs) in answering osteoporosis-related frequently asked questions (FAQs) derived from international osteoporosis-related associations. A standardized question bank was constructed based on FAQs summarized from three international osteoporosis-related associations. Six LLMs were prompted to generate responses to all questions under uniform conditions. Two osteoporosis experts independently and blindly evaluated all responses using 5-point Likert scales for accuracy and comprehensiveness. Inter-rater reliability was assessed using Cohen's κ coefficient. Nonparametric statistical analyses, including Kruskal-Wallis tests with Dunn's post hoc comparisons and Wilcoxon signed-rank tests, were performed to compare models' performance. A total of 528 responses were generated from 88 association-based questions. Inter-rater agreement was good for both accuracy and comprehensiveness (κ = 0.668 and 0.702, respectively). Significant differences were observed among models, with Claude Sonnet 4.5 demonstrating the highest overall performance on association-based FAQs. Although most models showed solid medical knowledge and basic reasoning ability, substantial variability was observed across question types and prompt formulations. Current LLMs demonstrate promising potential as auxiliary tools for osteoporosis-related health management. However, it is necessary to optimize the strategies for integrating medical knowledge and designing prompts, so as to better improve the accessibility of its clinical application.
24 May 2026
Read appraisal →Medicine
Tech-based evaluation of healthcare quality during the COVID-19 pandemic
The coronavirus disease 2019 pandemic profoundly disrupted global health systems, exacerbating preexisting inequalities in access to and quality of care. Disparities in service accessibility, continuity, coordination, and comprehensiveness highlight the need for a multidimensional, data-driven approach to healthcare evaluation. We conducted a retrospective analysis using 52,490 responses from the 2021 Global Burden of Disease COVID-19 Health Service Disruption Survey. We developed a 4-dimensional evaluation framework aligned with World Health Organization principles for people-centered care. Following variable selection, standardized preprocessing, and manual labeling, we trained and assessed 6 machine learning (ML) algorithms. We further optimized the 2 best-performing models - support vector machine (SVM) and random forest (RF) - using interpolation-based data augmentation, cross-validated hyperparameter tuning, and particle swarm optimization. SVM showed the highest baseline performance (accuracy 0.93), whereas RF achieved competitive macro-F1 scores. Interpolation-based data augmentation improved the generalizability of the SVM model, increasing its accuracy to 0.96. The RF model benefited most from particle swarm optimization-based hyperparameter optimization. Model predictions revealed persistent inequalities in healthcare service quality: respondents with lower income, lower educational attainment, and rural residence were disproportionately classified as receiving lower-quality services, whereas urban residence and higher income were associated with higher-quality care. Our scalable ML framework successfully assessed healthcare service quality during the pandemic, with optimized SVM and RF models demonstrating robust performance on high-dimensional, imbalanced survey data. Post-pandemic recovery strategies must prioritize socioeconomically disadvantaged populations. Furthermore, integrating ML tools can enhance real-time quality monitoring within health systems.
24 May 2026
Read appraisal →BMJ health & care informatics
Beyond the digital health tsunami: reaffirming biomedical and health informatics as the scientific foundation of digital healthcare transformation
Digital health (DH) has emerged as the dominant terminology for technology-enabled healthcare, reshaping institutional discourse, educational programmes and policy agendas. DH is often portrayed as a novel discipline that could supersede Biomedical and Health Informatics (BMHI), creating conceptual ambiguity and the risk of weakening scientific rigour in the design, evaluation and governance of digital interventions. This paper clarifies BMHI's disciplinary identity, characterises DH as a practice-oriented and implementation-oriented domain, and establishes the epistemic relationship between the two in order to inform academic strategy, curricula and professional identity. A comparative conceptual analysis examined scope, theoretical foundations, methodological approaches, educational frameworks, professional structures and stakeholder ecosystems in both domains, informed by recent debates on degree programmes and workforce training. We found that BMHI is a mature scientific discipline with explicit theoretical foundations encompassing data representation, knowledge modelling, clinical reasoning, decision support, interoperability, sociotechnical analysis, evaluation science, safety, ethics and governance. DH, by contrast, represents an implementation-oriented translational domain that operationalises BMHI through products, services and organisational transformation (eg, telehealth, mobile applications, artificial intelligence-enabled clinical services). While DH has amplified visibility and accelerated adoption, it lacks autonomous theoretical foundations and depends on BMHI for scientific validity, safety and sustainability. Three evolutionary scenarios are examined: DH absorbing BMHI (epistemically incoherent), convergence as parallel disciplines (conceptually flawed), and reintegration of DH within BMHI as its applied layer (scientifically coherent and strategically sustainable) but requiring disciplinary communication, curricular alignment and institutional leadership explicitly positioning DH as a translational layer of BMHI.
24 May 2026
Read appraisal →Studies in health technology and informatics
Interoperability Skills for Digital Health: A Reference Model from the XiA Project
As healthcare systems advance through digital transformation, achieving interoperability across Health Information Systems (HIS) relies on skilled professionals who enable effective integration and data exchange. Under the scope of the Xpanding Innovative Alliance (XiA) project, this article introduces a Reference Model for Interoperability Skills that defines the core skills needed by professionals to support interoperable digital health ecosystems. Through a systematic literature review, the model provides a structured view of competencies across professional profiles, grounded in international best practices. The proposed model serves as a strategic tool for workforce development, curriculum design, and organizational capability assessment. By promoting a shared understanding of interoperability, it supports the creation of a skilled, adaptable, and collaborative workforce capable of advancing integrated, data-driven healthcare across Europe.
24 May 2026
Read appraisal →Studies in health technology and informatics
Leveraging Large Language Models with Retrieval-Augmented Generation for Semantic Mapping of Clinical Data Lakes to SNOMED CT
Mapping local clinical concepts to standardized terminologies such as SNOMED CT is essential for semantic interoperability and large-scale research, but manual mapping is labor-intensive and difficult to scale. We report a preliminary evaluation of a hybrid approach combining retrieval-augmented generation (RAG) and a large language model (LLM) to support the mapping of hospital datalake concepts to SNOMED CT. Using a dataset of 2,768 concepts annotated by the SIMED across five semantic categories (organisms, healthcare locations, laboratory samples, allergies, and clinical assessments), the pipeline consistently outperformed retrieval-only baselines. For organism concepts, strict accuracy reached 92.8% (weighted 98.0%, top-3 96.7%), while allergy and assessment concepts achieved 84.7% and 72.1% strict accuracy, respectively. Healthcare location and laboratory sample concepts remained more challenging, with strict accuracies of 57.5% and 57.1%, and weighted accuracy varied strongly depending on frequency distributions. Overall, strict accuracy was 72.8%, weighted accuracy 76.1%, and top-3 accuracy 83.1%. These findings suggest that combining RAG with LLM-based reasoning can reduce expert workload while maintaining accuracy, representing a step toward scalable, AI-assisted semantic interoperability in healthcare. The approach relies on a retrieval-augmented generation pipeline using embedding-based candidate retrieval followed by LLM-based disambiguation (GPT-4.1-mini), and is evaluated against a retrieval-only baseline. By improving semantic consistency in clinical data lakes, it indirectly supports interoperable access to patient-level health data.
24 May 2026
Read appraisal →Studies in health technology and informatics
A Federated Benchmark for Clinical Natural Language Processing (FedDRAGON)
We introduce the FedDRAGON challenge, a federated learning benchmark for clinical natural language processing. The challenge includes 12 information extraction tasks where data is extracted from clinical reports from 4 Dutch care centers. Baseline results show that the performance of the federated models surpass single-center performance and approaches that of centralized models. Benchmark, code, and pre-trained LLMs are publicly available.
24 May 2026
Read appraisal →Studies in health technology and informatics
Evaluating the Potential of Machine Learning for Discharge Management on Routine Health Insurance Data
Machine learning (ML) has great potential in healthcare, especially with large structured data. Routine health insurance claims (HIC) data are a valuable resource, comprising standardized longitudinal patient information. However, to fully leverage ML in HIC, challenges such as large data volumes, variability, data gaps, sparsity, and high dimensionality must be addressed. To overcome these barriers, our publicly funded KI-THRUST project accessed a large-scale dataset of 2 million anonymized routine insurance records from four German providers. Subsequently, a thorough performance evaluation of various linear and nonlinear ML models (logistic regression, AdaBoost, Random Forest, Neural Network) for their potential to predict clinical outcomes such as mortality and unplanned readmissions was conducted, and the results were interpreted using explainable AI methods (Integrated Gradients, Shapley values, LIME). In particular, we investigated the predictive value of feature sets comprising base demographics (M1) and ICD-10 codes for previous conditions and procedures (M2, M3). While AdaBoost achieved the best overall performance, the less complex logistic regression performed comparably. The additional diagnostic and procedural information in M2 led to a significant performance gain, whereas additional time-related information did not. Overall, our results show the potential of large-scale HIC data when combined with ML methods and XAI. Ultimately, paving the way for future applications like discharge management systems and other healthcare areas.
24 May 2026
Read appraisal →Studies in health technology and informatics
Large Language Models for Automating Conformance to Health-Data Standards: The Interoperability Case of HL7 FHIR and OMOP
Interoperability standards enable different healthcare systems to exchange electronic health information seamlessly, and support large-scale analytics needed for current Secondary uses of healthcare data. Transforming diverse clinical inputs into standards-compliant, analyzable data remains labor-intensive and error-prone. Moreover, the learning curve for their proper exploitation can be steep due to extensive documentation, specific technical requirements, and the need for both semantic and technical integration, making them not fully accessible to practitioners in the healthcare domain unless they undergo extensive training. Our study investigated whether large language models can automate conformance while preserving syntactic validity, semantic correctness, and reproducibility, thereby simplifying the adoption of these standards. Using diabetes measurements as a probe domain, we validated extraction fidelity, coding accuracy against controlled vocabularies, and conformance to the domain's main standards. Our prototype emits FHIR Observations and OMOP MEASUREMENT rows.
24 May 2026
Read appraisal →Studies in health technology and informatics
De-Identification of Free-Text Medical Records Using Large Language Models
This study evaluates large language models (LLMs) for automated de-identification of free-text medical records. Using 150 synthetically generated, personal health information (PHI)-enriched doctor's letters (23.2 PHIs per letter), we compared the full LLaMA-3.1-8B model (BF16, GPU) with a quantized variant (Q8, CPU) under zero-shot and few-shot prompting. Few-shot prompting improved performance from a macro F1 of 0.989 to 0.996 (recall = 0.994); the quantized Q8 model achieved a macro F1 of 0.992 (recall = 0.987) while enabling fully local, GPU-free inference. Although LLMs perform at a level comparable to human annotation, false negatives remain a concern for highly sensitive PHI, such as names. While human annotators typically require 13.4 seconds per PHI, the LLMs de-identified complete letters within seconds, demonstrating an efficiency and scalability advantage alongside GDPR-compliant, high-precision de-identification.
24 May 2026
Read appraisal →Studies in health technology and informatics
Anonymization of Electronic Health Records by the Use of Dimensionality Reduction Techniques
Electronic health records (EHR) mostly consist of highly sensitive data, whose sharing is highly regulated due to privacy and security concerns. To address privacy protections, de-identification and anonymization techniques offer a more secure way of data transmission to ensure patient's privacy. Principal components analysis (PCA) has emerged as an anonymization technique. The applicability of more diverse and complex dimensionality reduction (DR) techniques such as Factor Analysis of Mixed Data (FAMD), or Autoencoder (AE), remains largely unexplored. The objective of this study is an investigation into the potential of FAMD and AE as anonymizers for EHR. The goal of this study is to apply advanced DR methods on three EHR datasets of varying sizes. Subsequently, a supervised prediction task is performed on the anonymized EHR using an artificial neural network (ANN), and the performance of the prediction task is compared to the performance of the original, non-anonymized EHR. The findings indicate that the use of FAMD and AE in the anonymization of EHR results in a comparable performance to that of the original, non-anonymized EHR in a downstream prediction task. This study thus indicates the potential of advanced DR as an anonymization technique for EHR, thereby underscoring the necessity for further investigation.
24 May 2026
Read appraisal →Studies in health technology and informatics
Accessibility of Digital Health and Welfare Services Among Immigrants in Finland
Digital health and welfare services support the delivery of the services. However, the health benefits are mixed at the population level, and further research is needed to better understand the driving factors of healthcare inequalities, particularly among vulnerable subgroups, e.g., immigrants. This qualitative study aimed to understand equality in service accessibility among immigrants, especially from the informational, experiential, and institutional perspectives. Five focus group interviews were conducted for immigrants. Data was analysed with deductive content analysis. Barriers for informational accessibility included linguistic challenges and inadequate digital literacy (DL); for experiential accessibility, usability challenges and disrespectful phone services; for institutional accessibility, challenges with electronic authentication identification for personalized digital services. Facilitators for the use of digital health and welfare services included linguistic improvements of services and support for DL. Further research, including interventions with participatory design and linguistic and digital support, is needed to improve accessibility of digital services for immigrants.
24 May 2026
Read appraisal →Studies in health technology and informatics
Mapping Existing Evidence on Physicians' and Patients' Experiences with GenAI in Clinical Communication and Documentation: A Rapid Review
GenAI tools like chatbots and ambient scribes are increasingly used in clinical documentation, raising ethical and safety concerns. This rapid review maps existing evidence on experiences and attitudes of physicians and patients toward using GenAI in clinical communication and documentation. Structured searches were conducted in PubMed and Web of Science. Studies were screened using PICO criteria. Data were extracted and summarized descriptively, with high-order thematic analysis of research aims and emerging recommendations. 41 studies were included: 75.6% focused on physicians, 9.8% examined patient perspectives exclusively, and 14.6% included both. Some of the themes identified were usability, exploration of perceptions and expectations, acceptability, and ethical implications. Recommendations emphasized the need for diverse cohorts, real-world validation, and ethical considerations. Current literature was skewed toward physician perspectives. Broader, meaningful inclusion of patients is needed to support equitable and trustworthy GenAI integration in healthcare.
24 May 2026
Read appraisal →Studies in health technology and informatics
Towards Clinical NLP in a Low-Resource Language: Structuring Estonian Clinical Text
Healthcare sectors generate large volumes of unstructured text, yet mature clinical NLP is lacking for low-resource languages. This project targets Estonian, a morphologically rich, low-resource language, and outlines an end-to-end pipeline that converts free text into interoperable, structured data. A design-science approach guides development of multilingual speech-to-text for clinical dictation, clinical NLP for entity extraction and context handling, terminology binding to the Estonian SNOMED CT extension, and export as HL7 FHIR resources with mappings to EHR fields. Expected benefits include reduced documentation time through dictation with automatic structuring, improved data quality and consistency via controlled terminology, and faster retrieval of clinically relevant information to support decision-making. Pilot evaluations with healthcare partners will assess accuracy, usability, and workflow impact.
24 May 2026
Read appraisal →Studies in health technology and informatics
Design and Evaluation of an Episodic Guideline-Driven Decision Support Engine
Clinical guidelines (GLs) standardize care but are complex to implement. Most clinical decision support systems (CDSSs) assume continuous use, which does not reflect real-world episodic workflows. We developed and evaluated e-Picard, a CDSS providing GL-based recommendations for episodic, on-demand consultations, enabling prospective decision support informed by retrospective quality assessment. At runtime, e-Picard analyzes offline patient data, computes fuzzy-logic-based compliance, identifies missed actions and generates context-specific recommendations. The system was applied to longitudinal geriatric data managed under pressure ulcers (PU) and diabetes (DM) GLs. Manual technical validation using 3,110 PU and 12,538 DM data instances (43 PU and 82 DM patients) achieved ≥99% correctness and up to 98% completeness. Retrospective simulation on 1,000 patients per domain (57,860 PU and 100,940 DM data instances) demonstrated potential adherence improvements from 68%-69% to 89%-97% for PU and from 14%-15% to 60%-87% for DM, with higher consultation frequencies increasing compliance and reducing care variability. These results demonstrate that episodic CDSSs can deliver accurate, context-aware support even under intermittent use.
24 May 2026
Read appraisal →Studies in health technology and informatics
Build and Query Indexes of Clinical Documents with Easy-to-Reuse Pipelines
Electronic Health Records are a central source of healthcare data, containing structured data alongside unstructured clinical texts. The latter capture detailed reasoning, observations, treatment plans and clinical evolutions, which are crucial for phenotyping, and real-world evidence generation. Natural language processing enables the extraction, thus the subsequent use, of these crucial elements; however, these extractions remain one-off, study-specific efforts. This is detrimental as the extracted elements could be valuable for future research. We present medkit Seshat, an open-source Python pipeline that: (1) ingests free text, (2) recognizes relevant entities, (3) normalizes them with OMOP vocabularies, (4) builds an index that can either be searched by concept or by document. In addition, we share a flexible web UI to illustrate the interest of built indexes in terms of search, text analysis and export. Seshat aims at facilitating the reuse and adaptation of this prototypical pipeline to various purposes, with the main objective of enabling the secondary use of results of phenotyping campaigns.
24 May 2026
Read appraisal →Studies in health technology and informatics
Assessing AI-Based Decision Support in Early Sepsis and AKI Recognition
Sepsis and acute kidney injury (AKI) remain among the most critical conditions in acute care, associated with high morbidity and mortality. Early risk recognition is essential but often hampered by nonspecific symptoms. Recent studies have demonstrated that AI- and ML-based Clinical Decision Support Systems (CDSS) can enhance the early detection of sepsis and AKI and support clinical decision-making [1-3]. In May 2025, the University Hospital Augsburg (UKA) integrated an externally developed AI-based CDSS to provide real-time risk predictions for sepsis and AKI. This study evaluates the effectiveness of the system in routine clinical practice using retrospective data from periods before and after its implementation. Primary endpoints include diagnostic accuracy. The results will indicate whether the CDSS contributes to improved early risk recognition and enhanced patient safety in a real-world hospital environment.
24 May 2026
Read appraisal →Studies in health technology and informatics
Electronic Health Record System Support for Physician-Patient Communication: National Cross-Sectional Survey on Outpatient Physicians' Experiences
The impact of digital patient engagement platforms (DPEPs) on interactions between physicians and patients has been widely studied from patients' perspectives. However, research into physicians' viewpoints remains scarce. This study aimed to identify electronic health record (EHR) characteristics that support physician-patient collaboration from the perspective of physicians working in outpatient care. A national cross-sectional survey of physicians' experiences with EHRs was conducted in Finland in 2021 where EHR coverage had reached 100% already in 2010. We selected the responses of 1983 physicians working in public sector outpatient clinics (primary and specialized care). Linear regression analysis was used to examine the relationship between eight EHR characteristics and perceived support for physician-patient collaboration and information exchange. Of these, rapid system responsiveness, ease of utilizing patient-provided data, support for adherence to treatment recommendations, assistance in preventing medication errors, and comprehensive and reliable patient data were found to be associated with support. The model accounted for 27% of the variance in the dependent variable. In addition to enhancing care outcomes, clinical decision support and easy access to patient data can aid physicians in interactions with patients. Further investigations should continue to analyze how these identified features could better support collaboration with patients - beyond DPEPs and visible interactions during appointments. Considering the rapid evolution of DPEPs, future studies should also take into account the depth of the DPEP-EHR integration.
24 May 2026
Read appraisal →Studies in health technology and informatics
Digital GO: A Clinical Decision-Support Dashboard for Gynecologic Oncology Tumor Board
Multidisciplinary Tumor Board (MTB) meetings are essential for coordinated oncologic decision-making but are often hindered by the complexity and fragmentation of patient data. This study presents the development of an interactive digital dashboard designed to streamline data access and visualization during ovarian cancer MTB sessions. The system leverages structured knowledge extracted from unstructured clinical reports using in-house large language model (LLM) pipelines. These outputs are harmonized within a cloud environment through dedicated processing workflows and mapped into a domain ontology, to ensure interoperability. The resulting dashboard, implemented using a dedicated functional module, enables both cohort-level exploration and patient-specific visualization, including longitudinal event timelines. Preliminary results demonstrate the feasibility of integrating LLM-based data extraction with a scalable visualization framework, offering a step toward the digital transformation of multidisciplinary oncology workflows.
24 May 2026
Read appraisal →Studies in health technology and informatics
Exploring a Large Language Model-Based Chatbot Use in Data Analysis: A Case Study of the Problems Related to the Do Not Attempt Resuscitation Order
The interpretative framework was used to explore how healthcare professionals (HCPs), artificial intelligence (AI), and researchers construct and negotiate meaning, trust, and authority in critical, highly ethical medical contexts. This case study aimed to explore and test the integration of the Microsoft Copilot® large language model (LLM) chatbot into ethically sensitive clinical decision-making (DM) to identify the problems HCPs face in do-not-attempt-resuscitation (DNAR) order-making and key areas for development. A sample of 100 HCPs' views was analyzed using qualitative content analysis. AI identified 15 thematic categories emerging from responses that described issues related to DNAR protocols and DM. The frequency of opinions in categories ranged from 4 to 16 (Copilot®) and from 2 to 16 (researchers). Based on a three-level frequency scale, AI's interpretations were especially evident in categories describing the roles of patients and families as well as practical issues related to DNAR orders. In contrast, the researchers focused on the importance of communication and documentation.
24 May 2026
Read appraisal →Studies in health technology and informatics
Frameworks for Digital Health Engagement: A Scoping Review Protocol
Digital health technologies (DHTs) are increasingly integral to healthcare delivery, yet sustained engagement remains inconsistent across users and settings. This scoping review aims to map frameworks and instruments that conceptualise digital health engagement across adoption and sustained use. Searches were conducted in Scopus, Web of Science, ACM Digital Library, and PsycINFO. The review will clarify conceptual coverage and identify gaps in digital engagement.
24 May 2026
Read appraisal →Studies in health technology and informatics
Can NLP Detect Loneliness in Electronic Health Records? A Proof-of-Concept Study
Loneliness is clinically important but under-documented in electronic health records (EHRs), posing challenges for secondary use and computational phenotyping. This study evaluated whether natural language processing (NLP) methods can detect and classify loneliness severity from clinical notes. Patients with a loneliness survey (mild, moderate, severe) were identified, and notes within six months prior to the survey were retrieved. An expert-expanded lexicon was applied, and transformer models (RoBERTa, ClinicalBERT, Longformer) were fine-tuned for loneliness severity classification. Large language model-based summarization of social and psychiatric history was also tested as an alternative input representation. Performance was evaluated using accuracy, weighted-F1, and per-class F1. All models achieved modest accuracy (0.3 to 0.7), and struggled to identify severe loneliness, reflecting sparse and inconsistent documentation even among surveyed patients. While summarization marginally improved accuracy, gains primarily reflected mild predictions. Manual review of 100 social worker notes from severely lonely patients found explicit mentions of loneliness in only two cases, confirming that relevant documentation is exceedingly rare. These findings demonstrate that model performance is constrained by the sparse and inconsistent documentation of loneliness in EHRs, rather than by deficiencies in the modeling approach itself.
24 May 2026
Read appraisal →Studies in health technology and informatics
MAP-CARE: Enhancing Cross-Lingual Medical Intervention Terms
Traditional keyword-based or single-language systems are not able to align data from surgical and interventional procedures, especially from non-English healthcare systems. This study aims to develop a pipeline for cross-lingual retrieval and integration of medical procedures data. Methods: MAP-CARE is a novel framework that leverages Large Language Models (LLMs) for translating and transforming medical procedures into a unified multilingual embedding space. Semantic embeddings are used to enhance retrieval accuracy and interoperability across languages and healthcare systems. Results: MAP-CARE demonstrated high accuracy in the translation and mapping of clinical terms. Its cross-language translation performance proved robust, achieving up to 90% accuracy in translating procedure classification codes across English, German, French, and Italian-when considering the correct term among the top five retrieved. Conclusion / Discussion: MAP-CARE offers a flexible, scalable, and robust solution for the multilingual and cross-system integration of medical procedural data.
24 May 2026
Read appraisal →Studies in health technology and informatics
Usability of Mobile Health Apps from the Patient Perspective: A Systematic Review
Mobile Health (mHealth) apps are widely used to support care, but patient-focused usability evidence remains fragmented. We conducted a systematic review (January 2012-September 2025) of PubMed, IEEE Xplore, and Scopus to identify studies assessing usability from the patient perspective using standardized questionnaires or structured interviews. Of 490 records retrieved, 146 studies met the inclusion criteria. Fifteen disease categories were represented, most commonly mental health, neurology, metabolic/endocrine, and oncology. Overall, 24 distinct usability assessment methods were reported, and most studies relied on a single method. Among standardized instruments, the System Usability Scale (SUS) was most common (50 studies, 34.3%); the mean SUS score across these studies was 76.37, above the acceptability benchmark of 68. Overall, mHealth apps were generally rated as usable by patients, but future work should adopt mHealth-specific standardized measures, enroll larger samples, and report methods transparently to enable benchmarking and inform better app design.
24 May 2026
Read appraisal →Studies in health technology and informatics
Large Language Models for Health Knowledge Modelling in Data Interoperability: A Scoping Review of Methods, Standards, and Applications
Achieving interoperability in healthcare requires semantic alignment of data with standards such as SNOMED CT, LOINC, and FHIR. Manual ontology mapping is slow and inconsistent. Evolving Large Language Models (LLMs) can automate knowledge modelling and data harmonisation. This review explores how LLMs are applied to enhance or automate health knowledge modelling for data interoperability. Following PRISMA-ScR guidance, PubMed and Compendex were searched. Studies applying LLMs to ontology alignment, terminology mapping, schema integration, or knowledge-graph construction were included. From 166 records, 20 studies met the inclusion criteria. Most used GPT-4 or BERT-derived models with retrieval-augmented or embedding-based methods. Reported gains included automation, improved mapping precision, and scalability across standards such as FHIR, OMOP, and UMLS. LLMs show strong potential for automating semantic interoperability in health informatics. Standardised evaluation benchmarks and explainable frameworks are essential for trustworthy adoption.
24 May 2026
Read appraisal →Studies in health technology and informatics
Long Covid and Digital Health in Switzerland: Potentials and Limitations
Post-COVID-19 (or Long Covid) presents a complex and persistent health challenge affecting a growing number of individuals worldwide. Digital health interventions (DHIs) offer potential support for symptom monitoring, self-management, and information access. However, their availability, quality, and sustainability remain uncertain, particularly in smaller healthcare markets such as Switzerland. This study provides a descriptive assessment of ten representative DHIs for Long Covid available from Switzerland. Each tool was evaluated using a 19-item framework derived from validated instruments. Three independent reviewers tested all interventions on iOS and Android devices, with consensus ratings across five quality categories. Results showed strong usability (average 6.3/8) and practical usefulness (5/8), but consistent weaknesses in accessibility (3.7/8) and data protection (4.6/8). Total scores ranged from 14.7 to 26.3 out of 38. Several Swiss-based apps were discontinued during the study period, highlighting market fragility. Findings underscore the need for inclusive design, transparent data practices, and institutional support if DHIs are to complement Long Covid care sustainably.
24 May 2026
Read appraisal →Studies in health technology and informatics
Hybrid AI Decision Support in Healthcare: Insights into Usability and Trust
Decision Support Systems (DSS) increasingly integrate Artificial Intelligence (AI) to enhance clinical reasoning, yet adoption depends on professionals' trust and usability. This study explored how hybrid AI decision support, combining rule-based guideline logic with generative reasoning, can be designed and evaluated in practice. Using a human-centred design approach within preventive child healthcare, a prototype integrating Large Language Models (LLMs) into an existing guideline-based DSS was developed and tested. Usability and trust were evaluated with nine healthcare professionals through task observations, usability scales, and interviews. The system showed good usability and moderate trust; highest scores were for traceability, workflow integration, and autonomy, while confidence indicators and contextual reasoning needed improvement. The study demonstrates the feasibility and acceptability of hybrid AI DSS in child healthcare and provides initial design requirements for usable, trustworthy, and human-centred AI integration across clinical domains.
24 May 2026
Read appraisal →Studies in health technology and informatics
Digital Health Tool Usability and Redesign: Preliminary Scoping Review Findings
Few usability or user experience studies report how insights resulted in improved digital health tool (DHT) design. Therefore, this paper reports on preliminary findings of a scoping review that examines this topic. A systematic search in five databases led to the inclusion of 102 studies that met the inclusion criteria, namely those reporting both usability evaluation and redesign outcomes. Categorisation of these studies identified four user groups: patients, healthcare professionals, public health users, and mixed groups. Many DTHs were web- (36%) or mobile-based (32%) applications, and most often newly developed (68%).
24 May 2026
Read appraisal →Studies in health technology and informatics
Training Future Health and Information Technologies Professionals on the European Health Data Space.
The European Health Data Space (EHDS) furthers the EU's digital health transformation by enabling secure access to electronic health data by facilitating environments for health data sharing. The topic of EHDS is relatively new and remains underrepresented in most academic curricula in both medical and informatics fields. This study evaluates the impact of a structured educational intervention to improve awareness, understanding, and preparedness among future health and information technology professionals to engage with EHDS-related systems and policies. Conducted at the University of Oulu, the intervention included training sessions, practical examples, and interactive discussions on topics such as the primary and secondary uses of health data, legal frameworks, and technical infrastructure. The survey's results showed significant gains in students' conceptual understanding of EHDS, particularly in distinguishing between primary and secondary uses of health data. We found that students with a European background reported slightly higher familiarity with EHDS-related rights and frameworks. The findings underscore the importance of embedding EHDS education in academic programs to build a digitally competent, informed workforce capable of supporting EHDS implementation. This intervention contributes to the portfolio of learning materials, developed by the SUSA project to integrate digital health policy and governance into multidisciplinary higher education curricula across the EU.
24 May 2026
Read appraisal →Studies in health technology and informatics
Advancing Knowledge in Evaluating the Clinical Impact of Large Language Models for Clinical Text Summarization: A Narrative Review
Large language models (LLMs) are increasingly explored for clinical text summarization from electronic health records (EHRs), where they could reduce documentation burden. While promising, errors in generated summaries may propagate misinformation, misrepresent clinical facts, and ultimately pose risks to patient safety. This review examined how the clinical impact of LLM-based summarization has been evaluated across four dimensions: utility, failure, patient safety risks, and bias. Literature was retrieved from PubMed between June 2024 and September 2025. Of the 144 retrieved studies, 32 studies were included in the analysis. Failure analysis was the most frequently reported (n=16, 50%), focusing on inaccuracies, omissions, and hallucinations, though definitions and methods varied widely. Utility analysis (n=8, 25%) examined workflow efficiency, readability, and understandability. Bias analysis (n=7, 22%) considered gender, race, and stigmatizing language. Only one study explicitly conducted patient safety risk analysis (n=1, 3%), using risk matrices. Findings indicate that evaluation frameworks are gradually moving beyond technical performance metrics; however, methodologies remain heterogeneous and lack standardization. Future research should establish consistent error taxonomies, structured risk assessment frameworks, and systematic bias detection methods to enable the safe clinical deployment of LLM-based summarization systems.
24 May 2026
Read appraisal →Studies in health technology and informatics
Machine Learning Applications Within the Earlier Medicine Framework for Stroke: A Scoping Review
This scoping review explores how machine learning (ML) has been applied to stroke research within the Earlier Medicine framework, which promotes proactive and personalized care through primary (preventive care), secondary (acute care), and tertiary (chronic care) prevention levels. Following PRISMA-ScR guidelines, studies were identified from PubMed and Scopus, yielding 105 eligible studies. Most studies focused on tertiary prevention, such as predicting in-hospital mortality, complications, and functional outcomes, while fewer addressed early risk prediction or acute-phase detection. Common predictors included age, NIHSS score, glucose, stroke volume, BMI, hypertension, and diabetes, reflecting reliance on routine clinical data. Traditional ML models (e.g., logistic regression, random forest, SVM) remain dominant, although Ensemble/Hybrid and Deep Learning models have increased steadily since 2021. ML demonstrated promising predictive accuracy and clinical potential, though challenges remain in data heterogeneity, model transparency, and external validation.
24 May 2026
Read appraisal →Studies in health technology and informatics
Electronic Patient Records, Real-Time Clinical Documentation and Burden: An Umbrella Review
This umbrella review synthesises evidence from reviews on clinicians' documentation burden (DB) associated with electronic patient records (EPRs) in hospital and primary care settings. Guided by Donabedian's model (structure, process, outcomes), it maps causes of burden, barriers to real-time documentation, and technological solutions. Following the Joanna Briggs Institute (JBI) methodology, searches across six databases identified 2,739 records, with 166 under full-text review. Findings of this review will inform strategies to optimise real-time EPR use while minimising DB.
24 May 2026
Read appraisal →JMIR human factors
Development and Psychometric Validation of a Usability Instrument Based on ISO 25010 for Electronic Health Record Systems in Peruvian Health Care Settings: Methodological Study
BACKGROUND: The usability of electronic health records influences patient safety, clinical efficiency, and the sustainable digital transformation of health care. While tools such as the System Usability Scale exist, few are based on structured software quality models or align with local regulatory frameworks, particularly in low- and middle-income countries. OBJECTIVE: This study aims to develop and psychometrically validate an instrument to assess electronic health record usability, based on the ISO 25010 (International Organization for Standardization) quality model, and in accordance with the Peruvian Technical Health Standard (Norma Técnica de Salud [NTS]). METHODS: A 3-stage methodological study was conducted following COSMIN (Consensus-based Standards for the Selection of Health Measurement Instruments) guidelines. First, a 23-item instrument was developed, grouped into 7 usability dimensions based on ISO 25010. Second, content validity was assessed using a modified Delphi method with 10 experts, using Aiken's V coefficient. Third, structural validity and reliability were evaluated in 115 participants (physicians, nurses, and administrative staff) using confirmatory factor analysis and Cronbach α. Additionally, the module's functional compliance with the NTS was verified. RESULTS: The instrument demonstrated excellent psychometric properties among 115 participants (53 physicians, 48 nurses, and 14 administrative staff); high content validity (Aiken's V=0.95; 95% CI 0.84-0.98), excellent internal consistency (α=.968), and optimal factor fit (Comparative Fit Index=1.000; Tucker-Lewis Index=1.001). Overall usability was perceived as very high by 94 of 115 users (81.7%), and high by 21 of 115 users (18.3%), with no perceptions at lower levels. The module met 28 of 35 requirements (80%) of the Peruvian Technical Health Standard. Administrative staff reported higher satisfaction (13/14, 93% rated it as very high) compared with clinical professionals (81/101, 80% rated it as very high), particularly in the dimensions of error protection and operability. CONCLUSIONS: This study provides the first fully psychometrically validated instrument in Spanish based on ISO 25010, System Usability Scale, and other theoretical foundations for evaluating the usability of information systems. The results highlight the importance of considering both international standards and local regulatory requirements (NTS) in the design of digital health systems. The instrument is applicable in Spanish-speaking contexts and can serve as a reference for future usability evaluations in the region as well as in low- and middle-income countries.
23 May 2026
Read appraisal →Journal of medical Internet research
Measuring eHealth Literacy in the European Economic Area, Switzerland, and the United Kingdom: Scoping Review
BACKGROUND: Digital tools continue to evolve and have the potential to improve health care delivery. However, they are associated with challenges, including accessibility issues and health misinformation. Individuals need eHealth literacy (eHL) to reliably use these tools, and providers require appropriate eHL measurement approaches to offer targeted solutions. For around 2 decades, researchers have been operationalizing and measuring eHL. OBJECTIVE: This paper aims to provide an up-to-date overview of how eHL has been assessed in recent years in the European Economic Area, Switzerland, and the United Kingdom and which methodological limitations need to be considered. METHODS: A scoping review was conducted. Records were searched via CINAHL, PubMed, and Google Scholar on January 31, 2025, and January 28, 2026. Peer-reviewed empirical papers published in German or English since 2020 that measured eHL in the European Economic Area, Switzerland, or the United Kingdom were included. The synthesis covered the publication trend, eHL measurement approaches and associated limitations reported in the included papers, eHL measurement frequency, countries and languages, and samples. RESULTS: In the final analysis, 132 papers published between 2020 and 2025 were included. The number of publications per year showed an overall upward trend (2020: 11/132, 8.33%; 2025: 35/132, 26.52%). Nine self-report eHL measurement instruments were used, the eHealth Literacy Scale being the most frequent (94/132, 71.21%). All included papers (132/132, 100%) reported data collection via surveys, and digital surveys were common (71/132, 53.79%). Reported limitations included potential self-report biases (37/132, 28.03%), selection biases due to the data collection modes (26/132, 19.70%), and limitations specific to 5 eHL measurement instruments (18/132, 13.64%). Most included papers (121/132, 91.67%) reported eHL results from a single measurement per participant. Data were collected in 22 countries within the target regions. The distribution of eHL measurement instruments varied considerably between countries. Data were collected in 22 languages within the target regions. Patients or individuals with health problems were a frequent target group (64/132, 48.48%). Most papers (104/132, 78.79%) described studies covering broad adult age ranges. Sample sizes ranged from ≤50 (7/132, 5.30%) to >2000 participants (8/132, 6.06%). CONCLUSIONS: To our knowledge, this is the first scoping review synthesizing eHL measurement limitations reported in empirical papers from the European Economic Area, Switzerland, and the United Kingdom. The identified limitations reported in the included papers potentially lead to biased results. Therefore, health care providers and researchers should take various factors into account when selecting eHL measurement approaches, such as eHL measurement purpose, target population, and data collection setting. Future research should address these constraints by adapting and developing new or revised eHL measurement instruments, including translated and culturally adapted versions. Policymakers should encourage health care providers to conduct methodologically well-founded eHL measurements as a basis for targeted solutions.
23 May 2026
Read appraisal →Nicotine & tobacco research : official journal of the Society for Research on Nicotine and Tobacco
Identifying Key Predictors of Smoking Cessation Success: Text-Based Feature Selection Using a Large Language Model
INTRODUCTION: The most effective way to reduce mortality and morbidity among current smokers is to quit smoking. Although about half of smokers attempted to quit, only one-tenth succeeded in 2022. Understanding key predictors of smoking cessation success would inform smoking cessation interventions and increase quitting rates. METHODS: We analyzed data from waves 5 and 6 of the Population Assessment of Tobacco and Health (PATH) study (December 2018 to November 2021). Using OpenAI's GPT-4.1, we identified the top 45 variables from wave 5 that are highly predictive of 12-month smoking abstinence in wave 6, based on descriptions of survey variables. We then validated the predictive power of the GPT-4.1-selected variables by comparing the performance of eXtreme Gradient Boosting (XGBoost) trained on different sets of variables. Finally, we derived insights into the top 10 variables, ranked according to their SHapley Additive exPlanations values. RESULTS: The performance of XGBoost trained with all possible wave 5 variables and the 45 selected variables was almost identical (AUC:0.749 vs AUC:0.752). The top 10 variables included past 30-day smoking frequency, minutes from waking up to smoking first cigarette, important people's views on tobacco use, prevalence of tobacco use among close associates, daily electronic nicotine product use, emotional dependence, and health harm concerns. CONCLUSIONS: The high predictive performance of XGBoost, when trained on the selected variables, underscores the efficiency and efficacy of GPT-4.1-based feature selection. The top 10 variables include various risk factors that have been previously reported in the literature for their influence on smoking behavior. IMPLICATIONS: Our findings do not establish causal relationships between the selected predictors and 12-month smoking abstinence. However, identifying these key predictors provides valuable insights into the factors highly associated with smoking cessation success. This study demonstrates the ability of OpenAI's GPT-4.1 to perform feature selection using only the textual descriptions of variables. The efficient and successful application of GPT-4.1 for variable selection highlights the potential of integrating artificial intelligence tools into tobacco research to guide resource-efficient and targeted intervention strategies.
23 May 2026
Read appraisal →PloS one
Audit data governance for disability-inclusive public services: A systematic review and integrative S-A-C framework
This study consolidates interdisciplinary evidence on how audit data governance shapes accountability and service equity in data-intensive public services, with specific sensitivity to disability-inclusive delivery. Guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020, we conducted a protocol-driven systematic literature review of peer-reviewed studies and selected institutional and standards sources published between 2010 and 2025. We searched Scopus and complemented database retrieval with targeted searches of audit, digital government, accessibility, and data-governance portals. We applied a theory-led narrative synthesis to accommodate heterogeneity in designs, domains, and outcome reporting. Synthesizing 125 records, we find that governance effects are produced through three interacting dimensions-service delivery for people with disabilities (S), audit data governance (A), and data-driven culture (C)-that jointly condition transparency, contestability, and distributive impacts. Service delivery arrangements shape where data are generated, which encounters are recorded, and whether access needs, reasonable accommodations, and complaints become organizationally legible. Audit data governance conditions the credibility and usability of audit trails through provenance, metadata, retention, audit logging, and role-appropriate oversight access, shaping both oversight feasibility and privacy risk. Data-driven culture conditions how evidence is interpreted and acted upon, including data literacy, the standing of experiential evidence, and whether audit signals support learning or only compliance. Across the corpus, recurring tensions emerge between optimization and inclusion; transparency and privacy; standardization and reasonable accommodation; automated decisioning and meaningful contestability; and compliance-oriented auditing and learning-oriented governance. Drawing these mechanisms together, we propose an integrative framework linking S, A, and C to explain how auditability and disability inclusion are co-produced across the digital service lifecycle. The framework identifies leverage points for governance reform and evaluation and outlines implications for future empirical testing and for practitioners seeking trustworthy, oversight-ready, inclusive digital public services. A review protocol was prepared before the review; the review was not registered.
23 May 2026
Read appraisal →PloS one
Learning semantic similarity from sentence pairs using hybrid features centric approach and explainable siamese neural networks.
Semantic embeddings play an important role in modern natural language processing because they help models understand meaning beyond individual words. Accurate text similarity is essential for many applications such as search, automated scoring, summarization, and question answering. However, existing methods based on Term Frequency-Inverse Document Frequency (TF-IDF) or simple lexical overlap often fail when sentences differ in length, structure, or word choice. These methods are less in performance, especially when working with short or medium-length sentences where meaning is expressed in different ways. This study explores sentence-level similarity using a Siamese BiLSTM model that learns deep semantic patterns and relationships between two sentences. The model captures contextual meaning, word interactions, and paraphrastic variations more effectively than traditional approaches. Experimental results show that the proposed model achieves the highest performance among machine-learning regressors, with lower errors and improved stability. Compared to TF-IDF, cosine similarity, and feature-based regressors, the Siamese model provides more accurate judgments of semantic closeness with RMSE of 0.16 and MAE of 0.107. Feature-level analysis using TF-IDF, Jaccard similarity, and embedding distances further supports these findings. Explainable AI techniques (SHAP, LIME) confirm model transparency by highlighting meaningful semantic cues and distributing attention across important linguistic features.
23 May 2026
Read appraisal →PloS one
Enhancing aviation safety: An 80-year data-driven model for classification of aviation incident and accident
The aviation system is safety-critical by nature, and any occurrence of an incident or accident can lead to the loss of human life and significant operational disruptions. The International Civil Aviation Organization (ICAO) emphasizes that every flight must take off and land safely-a goal achieved over 126,000 times daily. Despite major advancements,mishaps and accidents continue to occur, underscoring the need for robust safety management systems. The accurate classification of aviation occurrences (Incident or Accident) reports is essential for safety management, yet manual review is time-consuming and prone to inconsistency. While incident/accident labels are assigned during reporting, automated classification enables rapid triage, detection of potential mislabeling, and support for severity assessment in high-volume aviation safety operations. To address this,we developed and compared three machine learning classifiers-Multinomial Naive Bayes, Random Forest, and Support Vector Machine-using TF-IDF vectorization on an 80-year dataset of 53,770 aviation occurrence summaries obtained from the Transportation Safety Board of Canada. A two-stage evaluation strategy was employed, consisting of an initial 80/20 train-test split to create an independent test set, followed by 5-fold cross-validation applied exclusively to the training data to ensure robustness and prevent optimistic bias. The Support Vector Machine (SVM) classifier achieved the highest classification performance, attaining an accuracy of 98.06% during 5-fold cross-validation, with consistent results across folds, demonstrating its effectiveness in managing high-dimensional textual data and dataset complexity. The proposed framework provides a robust foundation for automated aviation safety report processing, offering practical value for (1) early triage of safety reports, (2) identification of potentially mislabeled cases requiring expert review, and (3) integration into downstream severity assessment pipelines. This work advances beyond prior classification studies by establishing a benchmark on the largest historical aviation safety dataset while delivering a deployable and operationally relevant framework for real-world safety management applications. The findings offer valuable insights for regulatory authorities and airline operators, contributing to enhanced safety oversight, improved response strategies, and safer aviation operations.
23 May 2026
Read appraisal →JMIR formative research
MyLungHealth, a Patient-Facing Education Tool for Lung Cancer Screening: Qualitative User-Centered Design Study
BACKGROUND: Lung cancer remains the leading cause of cancer-related mortality worldwide, with low-dose computed tomography screening demonstrating an approximately 20% reduction in mortality among high-risk individuals. Despite this benefit, screening prevalence remains suboptimal, with often less than 20% of eligible individuals reported to be up to date on screening. Shared decision-making is essential for effective lung cancer screening (LCS) implementation, with decision aids shown to enhance patient knowledge and engagement. OBJECTIVE: The aim of this study is to identify patient preferences, concerns, and design considerations through qualitative evaluation of MyLungHealth, a personalized patient-facing educational tool for LCS integrated with electronic health records, and to describe how these findings informed iterative design modifications. METHODS: We employed qualitative research methods through focus groups (n=34) and individual interviews (n=18) with individuals who met screening eligibility criteria. Participants were recruited from the University of Utah Health and New York University Langone Health between May and December 2023. Feedback was analyzed using Braun and Clarke's thematic analysis principles. RESULTS: Six themes were organized into three overarching domains. Domain A included interpretation and impact of personalized risk information: theme 1, difficulties interpreting risk information, and theme 2, varied impacts of risk information on motivation. Domain B included autonomy, privacy, and user interface preferences: theme 3, desire for autonomy and control over personal health data, and theme 4, preference for straightforward language and multiple information formats. Domain C included integration with clinical workflows and patient portal systems: theme 5, expectations for integration with health care provider workflows, and theme 6, mixed experiences with personal health record systems. These insights led to key design modifications, including simplified risk presentation, multimodal content delivery options (video and text), and implementation of electronic health record alerts for clinicians. CONCLUSIONS: The user-centered design process for MyLungHealth revealed important considerations for developing effective patient education tools for LCS. The findings highlighted the need for simplified risk presentation, personalized information delivery, and integration with clinical workflows. These findings underscore the importance of balancing comprehensive risk communication with user accessibility.
23 May 2026
Read appraisal →ACS sensors
Artificial Intelligence-Driven Wearable Sensors for Cardiovascular Health Monitoring
Wearable sensing technologies hold great promise for continuous and non-invasive monitoring of cardiovascular health, offering new avenues for early detection and effective management of cardiovascular diseases. Despite these advances, several critical challenges remain, including inefficiencies in sensor design, the degradation of calibration model accuracy over time, and difficulties in processing complex physiological signals under real-world conditions. The integration of artificial intelligence (AI) into wearable cardiovascular monitoring systems presents transformative solutions to these limitations by improving system performance, adaptability, and clinical applicability. In this review, we provide an overview of current wearable technologies for cardiovascular monitoring, including their mechanisms, characteristics, and limitations. An in-depth discussion is subsequently presented on AI-driven wearable cardiovascular monitoring, encompassing five critical facets: sensor design, sensor calibration, signal processing, pattern recognition, and disease management. Furthermore, we analyze the challenges associated with the widespread use of AI-based wearable systems and provide insights into potential strategies for addressing these challenges. This review is expected to serve as a roadmap for future research and development in the rapidly evolving field of intelligent cardiovascular health monitoring.
23 May 2026
Read appraisal →JMIR formative research
Patients' mHealth Apps Usage and Data Privacy, Security, and Confidentiality Concerns: Exploratory Study
BACKGROUND: The Technology Adoption Model (TAM) offers a potential framework for elucidating the relationships between data privacy or security concerns and behavioral intention, perceived usefulness (PU), and perceived ease of use (PEOU) of mobile health (mHealth) apps, particularly for patients' self-care management. In Saudi Arabia, limited information is available on these pertinent research areas despite the government's relentless efforts to bolster the use of mHealth apps. OBJECTIVE: This study applies the TAM and the psychosociocultural framework to explore the influence of patients' data privacy and security concerns on the PU, PEOU, and behavioral intention to use mHealth apps for self-care management in Saudi Arabia. METHODS: A cross-sectional study was conducted by recruiting patients using mHealth apps for self-care from various provinces in Saudi Arabia. Research instruments were developed based on the components of 2 theories: the psychosociocultural framework and TAM, which were then piloted, validated, and distributed to participants via Google Forms. Linear regression models were performed to test the hypothesized relationships. RESULTS: Overall, 567 patients using mHealth apps participated in the study. Slightly more than one-third (217/567, 38.2%; range 35.6%-41.4%) of the participants expressed a high level of concern regarding data privacy, confidentiality, and security, with significant predictors being female gender, higher educational qualifications, and younger age groups (<46 years). About 18% to 25% of the variance in PU, PEOU, and behavioral intention to use mHealth apps was explained by the tested factors. Patients were more likely to have higher PU following a unit decrease in data confidentiality (β=.31; P=.01) and security concerns (β=.47; P=.01). The PEOU of mHealth apps increased as users demonstrated less concern regarding data privacy (β=.18; P=.001), confidentiality (β=.24; P<.001), and security (β=.43; P=.02). Likewise, behavioral intention to use mHealth apps also increased significantly following a reduction in respondents' concerns toward data privacy (β=.18; P=.02), confidentiality (β=.24; P=.03), and security issues (β=.36; P=.01). CONCLUSIONS: Specific demographic factors and concerns regarding data security and privacy influence patients' PU, PEOU, and behavioral intention to use mHealth apps for self-care management. Targeting the age-, education-, and gender-based differences regarding the usage of mHealth apps. Health care providers and policymakers may consider age-, education-, and gender-based differences when developing strategies to improve the adoption of mHealth apps among the Saudi patient population.
22 May 2026
Read appraisal →JMIR infodemiology
Japanese-Language AI Agent System for Human Papillomavirus Vaccine Infoveillance and Public Communication: Development and Feasibility Evaluation
BACKGROUND: Human papillomavirus (HPV) vaccine hesitancy remains a significant public health challenge in Japan, where proactive vaccination recommendations were suspended between 2013 and 2021. The resulting information gap between medical institutions and vaccine-hesitant populations is exacerbated by misinformation on social media platforms. Traditional public health communication strategies cannot address individual queries while simultaneously monitoring population-level discourse. OBJECTIVE: This study aimed to develop and conduct a feasibility evaluation of a dual-purpose artificial intelligence agent system that delivers verified HPV vaccine information to the public through a conversational interface while generating infoveillance reports for medical institutions based on user interactions and social media discourse. METHODS: We implemented a system with 3 components: a vector database integrating 139,803 documents, including academic papers, Japanese government sources, news media, and social media posts; a retrieval-augmented generation chatbot using a ReAct agent architecture with iterative multitool orchestration across 5 specialized knowledge sources; and an automated report generation system with modules for news analysis, research synthesis, social media sentiment analysis, including stance classification and topic modeling, and user interaction pattern identification. System performance was assessed using both automated and manual evaluation protocols on a scale from 0 to 5. RESULTS: The entire system functioned as expected. For single-turn evaluation, the chatbot achieved mean scores of 4.83 (SD 0.67; 95% CI 4.71-4.93) for relevance, 4.89 (SD 0.53; 95% CI 4.79-4.97) for routing, 4.50 (SD 1.29; 95% CI 4.27-4.70) for reference quality, 4.90 (SD 0.62; 95% CI 4.78-4.99) for correctness, and 4.88 (SD 0.54; 95% CI 4.78-4.96) for professional identity, with an overall mean of 4.80. Multiturn evaluation yielded higher mean scores: 4.94 for context memory (SD 0.25; 95% CI 4.84-5.00) and an overall mean of 4.98, with topic centering and identity achieving 5.00. The report generation system achieved high scores across all sections: 4.83 for completeness (SD 0.37; 95% CI 4.73-4.94), 4.88 for correctness (SD 0.33; 95% CI 4.77-4.96), and 4.12 for helpfulness (SD 0.48; 95% CI 3.98-4.27). Reference validity achieved perfect scores (5.00) across all periods, with citation correctness averaging 4.21 (SD 0.58; 95% CI 3.96-4.46). CONCLUSIONS: This feasibility study demonstrated that an integrated artificial intelligence agent system can support both public HPV vaccine communication and social media infoveillance in a Japanese-language context. Prospective deployment with real users is needed to assess actual public health impact.
22 May 2026
Read appraisal →JMIR medical informatics
A Practical Approach to Assessing the Completeness of Electronic Health Records for Medical Research: Data Quality Study
BACKGROUND: Data quality is the degree to which data are fit for their intended purpose and is described using quality dimensions. The increased use of medical data in clinical research and medical artificial intelligence development has rendered data quality assessment essential. Despite existing data quality definitions, frameworks, and tools, data quality assessment in real-world settings faces multiple challenges. This stems from a lack of understanding of how to assess real-world data quality and interpret the results. Therefore, practical approaches to data quality assessment are needed that are appropriate for diverse data environments, intended uses, quality dimensions, and requirements. OBJECTIVE: This study proposes a practical approach for assessing the completeness of electronic health records (EHRs) for medical research. This approach integrates structural completeness, rule-based assessment, and descriptive analyses of completeness and data diversity to clarify how data quality can be measured and meaningfully interpreted in practice. METHODS: The completeness of a large-scale EHR dataset from Gachon University Gil Medical Center was evaluated covering January 2005 to December 2023. Completeness was assessed using a three-part approach comprising (1) structural completeness assessment, (2) rule-based assessment, and (3) descriptive analyses of completeness and data diversity. Assessments were conducted using clinical data quality assessment tools. This practical approach was used to assess EHR completeness for medical research from 1,798,153 patient records. RESULTS: In the structural assessment, 12.8% (5/39) of the data tables were unavailable, indicating limited capturing of clinician free-text data. The rule-based assessment identified substantial missingness in vocabulary fields (38/124, 30.6%) and missing or special characteristic values in relation to observations (3,643,581/15,313,287, 23.8%), measurements (25,583,622/642,623,715, 4%), care sites (28/1715, 1.6%), and deaths (117/34,330, 0.3%). Descriptive analyses demonstrated a balanced gender distribution (886,489/1,798,153, 49.3% male and 911,664/1,798,153, 50.7% female) and a predominantly Korean racial distribution (1,739,628/1,798,153, 96.7%). Collectively, these findings illustrate the completeness quality of a multiperspective completeness assessment for medical research. CONCLUSIONS: This study demonstrates how data quality dimensions can be measured in practice through a real-world completeness assessment. This practical approach enables evaluation of EHR completeness and provides insights into data quality. Its findings have implications for researchers conducting data quality assessments and applying quality dimensions in medical research.
22 May 2026
Read appraisal →Postgraduate medical journal
Peer review in the age of artificial intelligence: a comparative study of human and AI-generated review reports
BACKGROUND: Peer review is central to maintaining scientific quality and helps editors make decisions. However, the volume of scientific publications continues to rise, placing pressure on the peer review system. With the rise of generative AI, its role in supporting peer review is gaining attention. This study aims to compare human-written and AI-generated peer review reports. METHODS: We analysed 398 peer review reports linked to 119 research articles published in BMJ Open in 2024. Publicly available reports and manuscripts were included. Editorials, corrections, and protocols were excluded. AI reports were generated using ChatGPT. All reports were anonymized and assessed by two independent reviewers. We conducted a hybrid thematic analysis. Frequencies of themes were calculated and compared by reviewer type. For quantitative comparison, we used the Mann-Whitney U test to assess differences in review quality scores and Fisher's exact test to compare the distribution of themes. All analyses were conducted using R software. RESULTS: Human reviewers gave more detailed and diverse comments. They addressed deeper issues like interpretation, originality, and applicability. AI reviews covered more sections but focused on routine or structural elements. AI outperformed slightly in format-related domains. Co-occurrence analysis showed human reviews linked diverse themes, while AI comments were structurally clustered. Shannon Index confirmed that human reviews were more thematically diverse. CONCLUSIONS: AI can support peer review by screening for basic errors. However, it lacks insight, critical judgment, and contextual awareness. Human input remains essential for meaningful review. Review-specific AI tools that preserve confidentiality are needed for future integration.
22 May 2026
Read appraisal →Journal of medical Internet research
Receipt of Medicines Information From the Internet and Other Information Sources Among Adult Medicine Users in Developed Economies, 2010-2025: Systematic Review
BACKGROUND: The internet, social media, and digital health tools have transformed access and receipt of medicines information (MI), complementing or replacing traditional sources, such as physicians, pharmacists, and package leaflets. OBJECTIVE: This study aimed to examine the prevalence and patterns of adult medicine users' receipt of MI from electronic sources (eg, internet and social media) compared with traditional sources, and to assess trends in the receipt of MI in developed economies since 2010. METHODS: A systematic search was conducted in CINAHL, the Cochrane Library, ProQuest, Scopus (including Embase and MEDLINE), and Web of Science databases for peer-reviewed studies conducted and published between January 1, 2010, and December 31, 2025. Studies including adults (≥18 years) in developed economies were eligible. The review followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines. Methodological quality and risk of bias were assessed using the Mixed Methods Appraisal Tool. Due to substantial heterogeneity, findings were synthesized narratively without meta-analysis. RESULTS: Twenty-six studies from 10 countries, involving 19,891 medicine users, were included. Most studies were cross-sectional surveys (n=19), with only one being a national long-term trend study. Physicians, pharmacists, and package leaflets consistently remained the most common MI sources, irrespective of patient group, country, or study design. No clear temporal trends in MI receipt were observed. Variations in the frequency of MI receipt were evident across research methods, medicine user groups, and types of medicines. Individuals with greater familiarity with the internet were more likely to receive MI from electronic sources, such as websites and social media. In most studies, MI obtained from social media was not distinguished from other internet-based sources. Half of the studies focused on heterogeneous medicine user groups (n=14), while the remainder examined specific medicine user groups (n=12). Although the methodological quality of the studies was generally acceptable, only 3 studies reported the use of an explicit theoretical framework. CONCLUSIONS: Traditional MI sources remain central for adult medicine users despite the growing role of electronic platforms. While the receipt of MI from electronic sources appears more common among internet-experienced users, no significant temporal trends in using these sources were identified. Further research is needed to better distinguish between different digital MI sources, including artificial intelligence-based MI sources, and to explore their evolving roles, implications for health-related decision-making, and associations with user characteristics and health contexts in the digital era. This review is innovative in systematically comparing traditional and electronic MI sources across developed economies and in highlighting methodological and conceptual gaps in existing research. By synthesizing cross-national evidence and identifying the need for clearer differentiation of digital MI subtypes, the study contributes to the field and informs the development of more targeted, reliable, and user-centered MI strategies in real-world health care settings.
22 May 2026
Read appraisal →JMIR formative research
Comparing Stakeholders' Perspectives on Parkinson Disease Management and Digital Technologies: Exploratory International Survey
BACKGROUND: Parkinson disease (PD) is a progressive neurodegenerative disorder that poses complex challenges for persons with PD, informal caregivers, and health care professionals. With growing interest in digital and predictive artificial intelligence (AI) tools for disease management, understanding the needs and digital readiness of these stakeholder groups is crucial. OBJECTIVE: This work aims to (1) identify digital practices for PD management among persons with PD, at-risk individuals, caregivers, and health care professionals; (2) compare these practices across groups; (3) explore stakeholder desires for AI-based tools; and (4) assess alignments and gaps to inform tailored AI solutions. METHODS: An anonymous cross-sectional online survey of an exploratory nature was distributed (from December 2024 to October 2025) in 5 languages and completed by 255 respondents. Descriptive statistics summarized responses to 41 questions, including stakeholder-specific items. χ2 tests were performed to examine stakeholder differences in desired AI features. RESULTS: Interest in predictive AI was high across stakeholder groups. Symptom tracking was the most desired feature (selected by more than 76% of the respondents), and personalized treatment recommendations came second for both persons with PD and health care professionals; however, stakeholder priorities diverged in other areas. Health care professionals rated improving patient and informal caregiver engagement as significantly more important than persons with PD did, χ21 (n=205)=34.78, P<.001, and Cramer V=0.41. Despite considerable interest, the reported use of digital tools was limited, as most persons with PD did not use symptom-tracking apps or wearables, nor were they currently monitoring their condition, although many expressed intentions to begin. CONCLUSIONS: While predictive AI tools were viewed positively across groups, there were significant gaps in stakeholder preferences, highlighting the importance of tailored, context-aware design. Early diagnosis was not prioritized by persons with PD or health care professionals, likely reflecting the complexity of diagnosing PD in the absence of disease-modifying therapies. Coupled with the emphasis placed on preventive lifestyle guidance by persons with PD and those at risk, this highlights the importance of actionability in AI-based monitoring and prediction. Such actionability may also enhance perceived relevance and uptake, given that reported interest in digital health tools and self-tracking exceeded actual use. These findings offer early-stage insight to guide the development of future AI-based solutions for PD.
21 May 2026
Read appraisal →JMIR formative research
Application of an AI-Based Pediatric Early Warning Score in the Pediatric Emergency Department: Cross-Sectional Study
BACKGROUND: Pediatric emergency departments see a high volume of patients. Given that children often cannot describe their condition and there is a shortage of nursing staff, it is essential to identify the early warning signs of adverse conditions among children as quickly as possible. Current targeted care needs to be improved. OBJECTIVE: This study aimed to investigate the application of an artificial intelligence (AI)-based version of the Pediatric Early Warning Score (PEWS) in a pediatric emergency observation unit, analyze the relationship between PEWS and disease severity, and assess its impact on the length of hospital stay and hospitalization costs after admission, thereby providing a reference for targeted nursing care. METHODS: We performed a retrospective study. A total of 1233 pediatric patients admitted via the pediatric emergency department of a tertiary specialty hospital in Guangzhou from September 2023 to March 2024 were included. The patients were divided according to whether they triggered a PEWS early warning into an early warning group (PEWS ≥1) and a non-early warning group (PEWS=0) during emergency observation. Length of stay and hospitalization costs were compared between the early warning group and the non-early warning group. Differences between groups were assessed using the Mann-Whitney U test. We performed multivariable logistic regression to discuss the association of resource use metrics and PEWS status, adjusted by age, sex, and disease category (respiratory, neurological, and hematologic). RESULTS: Of 1233 patients, 597 (48.4%) triggered the PEWS early warning (mean score 2.44, SD 1.41), and 636 (51.6%) did not. In the early warning group, 68 children were transferred to the intensive care unit, with a mean PEWS of 3.32 (SD 1.73). Compared with the non-early warning group, the early warning group had a longer hospital stay (z=-5.180; P<.001) and higher hospitalization costs (z=-6.500; P<.001), and the differences between groups were statistically significant (P<.001). Among the top 3 admission categories-respiratory, neurological, and hematologic diseases-children in the early warning group had significantly longer hospital stays and higher hospitalization costs (all P<.01). The β coefficient for length of hospital stay was 0.053 (SE 0.010; Wald χ²1=5.533; odds ratio 1.055, 95% CI 1.035-1.075), while the β coefficient for hospitalization costs was 0.001 (SE 0.000; Wald χ²1=6.075; odds ratio 1.001, 95% CI 1.001-1.001). CONCLUSIONS: Compared with the non-early warning group, the early warning group had significantly longer hospital stays and higher hospitalization costs (P<.001); similar patterns were observed within respiratory, neurological, and hematologic disease categories (all P<.01). These findings show differences between children who triggered the warning and children who did not, providing a reference for identifying critically ill children for targeted care.
21 May 2026
Read appraisal →PloS one
Optimizing bike-sharing station locations: A machine learning and artificial neural networks approach using geospatial and demographic data
In the modern world, public transportation amenities are noticeably on the rise, with urban bike-sharing systems becoming well-established in many major cities. However, not all cities have these systems, and planning optimal locations for bike-sharing stations is a complex task that requires consideration of many factors. To address this, the authors of this research paper developed a model to predict suitable locations for bike-sharing stations, utilizing machine learning techniques and artificial neural networks. These techniques utilized land cover and demographic data to train the model, achieving a high accuracy of 0.977. The predicted bike-sharing stations not only align with existing networks but also support their expansion, as many suggested locations are near major intersections and public transportation stops, confirming their suitability for the urban bike network. Additionally, the model was applied to Rzeszów, a city without a current bike-sharing system, where it successfully identified optimal locations for new stations. This demonstrates the methodology's practical applicability and its valuable support for planning bike-sharing infrastructure in urban areas.
21 May 2026
Read appraisal →Journal of medical Internet research
Alignment Between Cardiologists and AI-Driven Diagnostic Systems: Mixed Methods Study
BACKGROUND: The clinical value of artificial intelligence (AI)-based diagnostic systems depends not only on their accuracy but also on how well their outputs integrate with clinicians' judgments in practice. Critical knowledge gaps remain regarding diagnostic concordance between AI and clinicians in stress echocardiography interpretation, patient characteristics predicting discordance, and how cardiologists respond when AI recommendations conflict with their clinical diagnoses. OBJECTIVE: This study examined the diagnostic alignment between an AI-driven stress echocardiography system (EchoGo Pro [EGP]) and cardiologists' diagnoses of coronary artery disease (CAD), identified predictors of concordance and AI scan rejection, and explored cardiologists' decision-making strategies when disagreements arise. METHODS: We conducted mixed methods research. The quantitative study analyzed concordance between EGP and cardiologists using data from 854 participants with suspected CAD in the multicenter PROTEUS randomized controlled trial. Logistic regression identified predictors of agreement, disagreement, and scan rejection, adjusting for age, sex, smoking status, BMI, and cardiovascular risk factors (hypertension, hypercholesterolemia, diabetes, family history of CAD, and prior CAD events). To gain deeper insight into discordance, we conducted a qualitative study analyzing survey responses from 61 UK consultant cardiologists recruited via Qualtrics, exploring their perceptions of AI tools, the risks of following discordant AI recommendations, and their typical responses to AI-clinician disagreement. RESULTS: EGP and cardiologists agreed in 60% (512/854) of the cases, but agreement was significantly lower among patients with hypertension (OR 0.58, 95% CI 0.38-0.89; P=.01), diabetes (OR 0.56, 95% CI 0.35-0.90; P=.02), and pre-existing CAD (OR 0.48, 95% CI 0.30-0.77; P=.002). EGP rejected 26.1% (222/854) of the scans due to insufficient image quality, with rejection significantly more common in male patients (β=0.35; P=.03) and those with a family history of CAD. If a positive CAD diagnosis was assigned when either cardiologists or EGP identified CAD, the proportion of positive cases increased from 17.9% (153/854) to 22.1% (189/854), potentially identifying additional at-risk patients. Survey respondents (50/60, 85% male; 26/57, 46% aged 40-49 years; 39/61, 64% White) required 65% to 69% confidence in their initial diagnosis to justify disregarding contradictory AI recommendations. The survey findings revealed cardiologists treated AI recommendations as advisory rather than definitive. When facing discordance, they retained confidence in their judgment and sought corroboration through additional testing, data review, or second opinions rather than deferring to AI. Paradoxically, cardiologists with higher confidence in AI tools required greater confidence in their own diagnosis to disregard AI recommendations (β=7.73; P=.02). Cardiologists attributed discordance primarily to AI's inability to incorporate patient history, comorbidities, and broader clinical context. CONCLUSIONS: EGP shows promise as an adjunctive tool but struggles with multimorbid patients and exhibits high, uneven rejection rates. Cardiologists use AI to prompt scrutiny, not replace judgment. Future systems need to integrate wider patient data with imaging and minimize bias through representative training to avoid exacerbating inequities.
21 May 2026
Read appraisal →JMIR formative research
Development, Feasibility, Acceptability, and Usability of an Artificial Intelligence-Powered Chatbot (Suzy) to Support Patients in Substance Use Disorder Recovery: Multiphase Study
BACKGROUND: Substance use disorder (SUD) remains a major public health crisis in the United States, with significant challenges in treatment access, retention, and workforce capacity. SUD care teams, including addiction medicine physicians and peer recovery coaches (PRCs), support patients receiving SUD treatment but face heavy workloads and burnout. Artificial intelligence (AI) innovations, particularly large language model (LLM)-based chatbots, may extend PRC support and provide patients with on-demand recovery support between clinic visits and PRC contacts. However, evidence on their development, feasibility, acceptability, and usability in addiction services remains limited. OBJECTIVE: This study describes the development, feasibility, acceptability, and usability of an AI-powered health coaching chatbot (Suzy) designed to support patients in SUD recovery. METHODS: A total of 2 clinicians, 5 researchers, and 2 technology developers led a small, multiphase pilot study. In the formative phase, they conducted focus groups and qualitative in-depth interviews with 12 health care professionals and 8 patients with substance use histories to specify chatbot functions and develop a rule-based chatbot. In phase 2, they conducted usability testing of the rule-based chatbot with 8 patients who reported substance use and completed standardized tasks, surveys, and qualitative interviews. Measures included the System Usability Scale (SUS), Net Promoter Score (NPS), and Single Ease of Use Question (SEQ). In phase 3, they developed an LLM-based chatbot co-designed and fine-tuned with PRCs and other SUD experts. RESULTS: Rule-based chatbot functions included craving management, appointment reminders, resource referrals, care team contacts, and goal setting. Usability task testing supported feasibility. In this small pilot sample, quantitative and qualitative feedback indicated acceptability and usability, with an average SUS score of 93 (benchmark 68), an NPS of 63 (benchmark 35), and a mean SEQ score of 6.5/7. Patients valued Suzy's approachable, nonjudgmental language and features that promoted accountability, self-monitoring, and 24/7 availability, while emphasizing that chatbots should supplement but not replace human support. The LLM-based chatbot development emphasized information accuracy, safety escalation protocols to mitigate risks of inappropriate chatbot responses, human-in-the-loop features, and expanded conversational flexibility and personal tailoring. CONCLUSIONS: In this pilot study, a rule-based chatbot designed to support SUD care demonstrated feasibility, usability, and acceptability. LLM-based chatbot development required more robust safety and emergency reporting features, while offering more patient-responsive conversational functions. By providing on-demand coaching, referrals, and reminders, Suzy may extend the reach of care teams, alleviate provider burden, and enhance patient engagement. Additional work is needed to understand how to best integrate Suzy into patients' recovery journeys to ensure human support remains accessible and prioritized. LLM evaluation was based on expert testing and safety review. Clinical effectiveness, including the impact on substance use, was not evaluated. Next steps include evaluating the LLM chatbot in real-world settings with larger samples and assessing its efficacy in reducing substance use.
21 May 2026
Read appraisal →JMIR medical education
A Competency Framework for Medical AI Education: Mixed Methods Study
BACKGROUND: Although artificial intelligence (AI) is increasingly adopted in health care, clinicians face barriers, including insufficient understanding, limited trust, and challenges in interpreting AI outputs. Existing frameworks, such as the United Nations Educational, Scientific and Cultural Organization (UNESCO) AI competency framework, lack clinical specificity. Additionally, there remains limited evidence on framework-based training programs for medical professionals. OBJECTIVE: This study aimed to (1) develop a medical AI competency framework and (2) apply the framework to design an AI training program and pilot 1 module. METHODS: We conducted a mixed methods study comprising 2 phases. In Study 1, we developed a medical AI competency framework by integrating the UNESCO AI framework with the Miller pyramid model. The framework was refined through expert input from 24 stakeholders (6 hospital administrators, 8 medical professionals, and 10 university instructors). Expert responses were analyzed using deductive content analysis based on predefined codebooks. In study 2, we designed an AI training program based on the framework, and evaluated it using a 2-round Delphi process. Nine educators with expertise in instructional design, medical education, and AI participated in the Delphi study. Consensus in round 1 was defined as an IQR≤1, agreement score proportion >75%, and full-score frequency >49.23%; in round 2, consensus was defined as agreement score proportion >80%. A pilot workshop with 28 participants and 4 instructors assessed the feasibility of 1 module using self-reported measures of satisfaction, engagement, and confidence. RESULTS: A 6D 4-level medical AI competency framework was developed. Among the 24 experts, 19 (79.17%) mentioned competencies related to AI foundations, and 23 (95.83%) mentioned competencies related to application skills. The framework was translated into a 5-module training program, covering patient-centered and ethical AI, privacy and security, medical AI applications, bias and health equity, and generative AI in health care. Each module included 5 elements: content, learning goals, teaching activities, learning resources, and assessment. The Delphi process achieved complete consensus across all 25 elements of the training program. The pilot workshop indicated high participant satisfaction (mean 4.00, SD 0.52), good engagement (mean 3.80, SD 0.71-mean 4.05, SD 0.51), and moderate self-reported confidence (mean 3.63, SD 0.53). These findings suggest that the module was feasible, although outcomes should be interpreted cautiously given the self-reported and short-term nature of the evaluation. CONCLUSIONS: The framework provides a structured reference for AI training program design in medical education. The workshop findings provide preliminary support for the feasibility of the program's technical module, while also highlighting the need for broader and longer-term evaluations. Future work should expand the framework and training program to new regions and delivery formats (eg, semester-long courses and continuing medical education) and evaluate their long-term impact.
21 May 2026
Read appraisal →Health policy and planning
Partial pictures: what routine health data can and cannot tell us about the quality of maternal and neonatal health services in Tanzania
Measurement of contact coverage of maternal and newborn health services may overestimate the benefits of the care that is provided, because the quality of care is not captured. It is therefore important to better understand the opportunities and challenges of capturing elements of care quality in routine health information systems (RHIS). This study explored healthcare workers' (HCWs), health sector managers' and policymakers' perspectives on the value and limits of routine health data to understand the quality of maternal and neonatal health services in Tanzania. We conducted qualitative research during two periods to capture perspectives across facility, district, regional and national levels. In Mtwara region in 2023, we conducted ethnographic observations at two hospital labour wards and 29 in-depth interviews with healthcare workers, hospital leaders, and district/regional managers. In 2025, we carried out 17 additional interviews with regional managers in Mtwara and with key national-level stakeholders. Our findings demonstrate that Tanzania's RHIS provides a valuable but partial picture of quality of care for maternal and neonatal health services. Care processes-including both provision and experience of care-are captured only to a limited extent. Using boundary object theory, we highlight how the same health information must serve diverse stakeholder needs. While there are opportunities to integrate more quality-of-care indicators, standardized RHIS cannot be expected to comprehensively capture the multifaceted nature of care quality. For quality measurement to support meaningful local health service improvement, a flexible, bottom-up approach is essential. However, the current emphasis on top-down oversight acts as a barrier for local-level data use. Relevant and feasible measurement of the quality of facility-based care requires a fundamental shift in current RHIS priorities-from systems that extract data for those 'up there' to platforms that also create value for those who provide care.
21 May 2026
Read appraisal →Journal of clinical pathology
AI triage of duodenal biopsies improves workflow
AIMS: To develop, deploy and evaluate artificial intelligence (AI) for triaging duodenal biopsies within a National Health Service (NHS) histopathology laboratory, with the aim of improving reporting turnaround times for clinically significant diagnoses. METHODS: The pathway was developed in the UK in an NHS laboratory. Rule-based automation software was used to find all newly scanned duodenal biopsy slides. Those with case numbers ending in an odd number were exported for AI triage and if they had significant AI predicted abnormalities they were prioritised for reporting. The cases with even numbers followed the routine reporting pathway. RESULTS: 313 cases (517 duodenal slides) were processed by the routine pathway, and 329 cases (533 duodenal slides) were processed by the AI triage pathway. AI processing took about 70 s per slide. The AI classifier had a sensitivity and positive predictive value (PPV) as follows: normal small bowel: 99.6%, 95.7%; coeliac disease: 86.7%, 100%; gastric heterotopia: 84.6%, 95.7%; adenoma: 88.9%, 88.9%; adenocarcinoma: 50.0%, 100%. In the AI triage workstream coeliac disease, and non-neoplastic abnormalities as a group, were reported quicker than in the standard workstream (6 days vs 10 days and 7 days vs 10 days, respectively, both p<0.005), but neoplastic lesions were not reported quicker. The cost of deployment and operation was reasonable. CONCLUSIONS: An NHS histopathology laboratory successfully developed and implemented an AI-based triage system for duodenal biopsies, achieving high diagnostic accuracy and significantly improving turnaround times for coeliac disease and non-neoplastic abnormalities as a group. This study demonstrates the feasibility and clinical value of locally developed AI tools within routine diagnostic practice.
21 May 2026
Read appraisal →Scientific reports
Analysis, control, and forecasting the dynamics of SIRD models with saturated treatment and nonlinear incidence
This study addresses the challenge of managing complex epidemic dynamics exhibited by a novel Susceptible-Infected-Recovered-Deceased (SIRD) model. The introduced model incorporates a biologically and behaviorally mediated nonlinear incidence rate alongside a saturated treatment function, realistically capturing both physical transmission risks and the human behavioral responses during an outbreak. Our comprehensive analysis establishes the mathematical rigor of the system through positivity, boundedness, and a detailed investigation of equilibrium points and bifurcations (transcritical, saddle-node, and Hopf), which reveal the potential for complex, oscillatory, and chaotic epidemic patterns. To mitigate the disease burden, we apply Pontryagin's Maximum Principle to derive optimal time-dependent prevention and treatment control strategies. Furthermore, we introduce a highly efficient machine learning framework based on a Logistic-Map Reservoir Computer (LMRC) for accurate, real-time forecasting of these complex dynamics. Numerical simulations validate the theoretical findings and reveal the model's rich dynamical behaviors, including stable limit cycles and chaotic regimes driven by parameter variations. The integrated analytical, control, and machine learning approach provides a robust toolkit for public health policymakers, offering data-driven insights crucial for designing adaptive, cost-effective interventions that directly support the achievement of Sustainable Development Goal 3 (Good Health and Well-being).
21 May 2026
Read appraisal →BMJ open
Measurement of quality of stroke care with national electronic health records: a prospective cohort study during and after the COVID-19 pandemic
OBJECTIVES: To evaluate the value of linked electronic health records (EHRs) for measuring stroke care quality in England before and after the COVID-19 pandemic, focusing on metrics not routinely captured: stroke incidence, dispensing of secondary prevention medications and a proxy of disability-time spent at home after stroke ('home-time'). DESIGN: Prospective cohort study using national linked datasets. SETTING: England-wide health data linkage including the Sentinel Stroke National Audit Programme (SSNAP), primary and secondary care, dispensed medications and mortality records, accessed via National Health Service (NHS) England's Secure Data Environment. PARTICIPANTS: 425 675 adults with a first stroke between 1 January 2020 and 31 December 2023; data were available for 304 210 in primary care, 279 825 in hospital admissions, 220 470 in SSNAP and 59 465 in death records. MAIN OUTCOME MEASURES: Annual stroke incidence; first-year medication dispensing rates for antiplatelets, anticoagulants, antihypertensives and lipid-lowering agents (with a 1-month washout period) and home-time at 180 days post stroke. RESULTS: Stroke ascertainment was highest when combining all sources, with 10.8% of non-fatal ischaemic strokes recorded exclusively in primary care and 19.4% of fatal strokes identified solely through death records. Standardised annual stroke incidence rose from 227.6 (95% CI 226.1 to 229.0) to 244.8 (95% CI 243.4 to 246.3) per 100 000 over the study period including the COVID-19 pandemic. During the COVID-19 lockdown, non-fatal stroke recordings decreased while stroke-related deaths rose, indicating that recording quality was sensitive to shifts in healthcare-seeking behaviour during the pandemic. Among people with ischaemic stroke, 89.1% received an antiplatelet or anticoagulant, 44.5% an antihypertensive and 80.5% a lipid-lowering therapy. For haemorrhagic stroke, these proportions were: for anticoagulants 13.5%, antiplatelets 13.2%, antihypertensives 46.6% and lipid lowering 41.1%. Medication dispensing for stroke prevention declined with increasing age and comorbidity, but varied little by ethnicity, region or pandemic period. Mean home-time within 180 days of stroke was 166.6 (95% CI 166.4 to 166) days, decreasing with greater age (141.4 days for 90 years or older (95% CI 140.7 to 142.1)), deprivation (166.4 days (95% CI 166.1 to 166.6) for most deprived quintile) and stroke severity (137.4 days for National Institutes of Health Stroke Scale (NIHSS) score on arrival over 22 (95% CI 135.8 to 139.1)) and increasing with years from the COVID-19 pandemic 2023 (169.3 days (95% CI 169.0 to 169.5) vs 2020 164.4 days (95% CI 164.1 to 164.7)). CONCLUSIONS: Standardised stroke incidence increased significantly over the study period, highlighting a growing public health burden that persisted despite disruptions due to the pandemic although variation in case ascertainment and stroke coding practices was observed. While secondary prevention coverage for antiplatelets and lipids was high, lower rates of dispensing of antihypertensives, particularly in older and comorbid populations, potentially signal a target for improvement. Home-time represents a sensitive, person-centred outcome that exposes disparities linked to socioeconomic deprivation and clinical severity that can be used to enhance routine stroke audits. These findings justify the expansion of linked EHR infrastructure and the modernisation of governance frameworks to enable the longitudinal evaluation of care quality beyond the COVID-19 era.
20 May 2026
Read appraisal →JMIR medical informatics
Extracting Social Determinants of Health From Electronic Health Records: Development and Comparison of Rule-Based and Large Language Model Methods
BACKGROUND: Social determinants of health (SDoH) are critical drivers of health outcomes but are often underdocumented in structured electronic health record (EHR) data. Instead, SDoH are more commonly recorded in unstructured clinical notes, and unlocking this information could have far-reaching implications for advancing population health research and informing clinical decision-making. OBJECTIVE: This study develops and systematically evaluates cost-efficient methods for extracting SDoH information from unstructured clinical text using rule-based natural language processing (NLP) and large language model (LLM)-based approaches. METHODS: We constructed a gold-standard annotated corpus comprising clinical text segments from 171 patients in the Mass General Brigham Research Patient Data Registry, covering 7 SDoH domain categories and 23 subcategories. A rule-based system (RBS) was developed and evaluated alongside 7 OpenAI GPT models (GPT-4o, 4.1, 4.1-mini, o4-mini, GPT-5, GPT-5-mini, and o3) under zero-shot and few-shot settings using multiple prompting strategies. We additionally implemented late-fusion ensemble approaches that combined outputs from rule- and LLM-based methods. Performance was assessed using precision, recall, and F1-score, alongside qualitative error analysis. RESULTS: The RBS achieved high precision for SDoH domain categories (0.96) but substantially lower recall (0.68). GPT-based models consistently outperformed the RBS in overall recall and F1-scores. The best domain-level performance was observed for GPT-5 and GPT-5-mini in few-shot settings (F1-score=0.89), while o4-mini achieved the highest subcategory-level performance (F1-score=0.88). A late-fusion ensemble integrating RBS and GPT outputs further improved domain-level performance (F1-score=0.92), with balanced precision (0.93) and recall (0.90), but did not improve subcategory-level performance. CONCLUSIONS: Recent GPT models with advanced reasoning capabilities, including the newly released mini models (eg, o4-mini and GPT-5-mini), demonstrated strong performance for SDoH extraction without task-specific fine-tuning and consistently outperformed the rule-based NLP system. Integrating rule- and LLM-based methods via late fusion further enhanced domain-level extraction performance. Our results demonstrate a cost-efficient framework for the accurate identification of SDoH from clinical text, facilitating downstream population health research and clinical informatics applications.
20 May 2026
Read appraisal →Journal of medical Internet research
Co-Lifecycle Governance for Learning Medical AI: A Hybrid Convergence Framework for Adaptive Regulatory Oversight
Artificial intelligence (AI) in health care is increasingly defined not by static algorithms but by adaptive intelligence-systems that evolve over time through interactions with data, clinicians, and clinical environments. This adaptive capacity creates a structural mismatch with regulatory frameworks built for technologies whose behavior remains static. As AI models drift, recalibrate, or degrade in real-world contexts, they dissolve the linear boundaries between design, deployment, and clinical interpretation. These temporal, epistemic, and organizational frictions expose responsibility gaps that cannot be resolved through incremental modifications to legacy oversight structures. Regulators across major jurisdictions are beginning to respond to these challenges, though with differing orientations. The United States advances mechanisms for predictable adaptation, including Predetermined Change Control Plans, real-world evidence frameworks, and life cycle-oriented quality management reforms. The European Union emphasizes precautionary, rights-based governance through the European Union Artificial Intelligence Act (AI Act) and modernized liability rules. South Korea, operating within a hyperconnected digital health ecosystem, has introduced the Digital Medical Products Act (DMPA), one of the world's first comprehensive statutory frameworks for learning medical AI. Despite philosophical differences, these regulatory trajectories converge on a shared insight: learning AI systems cannot be governed by static rules or episodic evaluation. This viewpoint proposes Co-Lifecycle Governance as a conceptual framework to synchronize regulatory oversight with adaptive intelligence. Rather than treating oversight as a discrete event, Co-Lifecycle Governance frames regulation as a continuous, synchronized process grounded in 4 pillars: continuous validation, agile change management, proactive performance surveillance, and distributed accountability. Each pillar functions as a structural antidote to the responsibility frictions that arise when AI systems evolve faster than expectations surrounding them. Together, these pillars provide a governance grammar capable of supporting safe, iterative model improvement while maintaining system-level trust. Drawing from the strengths of US predictability, European Union accountability, and Korean scalability, this paper outlines a hybrid convergence pathway that synthesizes predictability, accountability, and operational feasibility. Learning AI will not wait for governance to catch up; oversight must evolve in lockstep with adaptive intelligence. Co-Lifecycle Governance offers a foundation for regulatory systems that not only regulate learning AI but also learn with it-at the speed at which adaptive intelligence actually changes.
20 May 2026
Read appraisal →PloS one
Implicit bias in safety-aligned large language models: A multi-faceted evaluation of clinical decision-making and health equity
BACKGROUND: Large language models are increasingly integrated into healthcare for clinical decision support and patient communication. Although these models can pass explicit social bias tests, they may retain implicit biases-latent associations between social groups and attributes-that could influence medical judgment. OBJECTIVE: To systematically evaluate the presence, magnitude, and behavioral impact of implicit biases in large language models within the medical domain across six high-stakes categories: gender, race, socioeconomic status, health conditions, religion, and healthcare systems. DESIGN: A descriptive cross-sectional study using a multi-faceted evaluation framework. SETTING(S): Computational analysis of 10 mainstream global large language models, including proprietary models (ChatGPT-4o, Gemini-2.0-Flash) and open-source models (DeepSeek-V3, Qwen3). METHODS: We constructed 24 medical bias datasets across six categories. Bias was assessed using three methods: (1) the Large Language Model Word Association Test, a prompt-based method for revealing implicit biases; (2) the Large Language Model Relative Decision Test, a strategy for detecting subtle discrimination in situational decision-making; (3) Paired-Prompt Analysis, used to examine whether implicit associations predict discriminatory decisions. RESULTS: All 10 models exhibited systematic implicit biases (Mean IAT Bias > 0) across all categories, with the strongest biases observed in Race (Mean = 0.61) and Socioeconomic Status (Mean = 0.56). Advanced reasoning capabilities (Chain-of-Thought) did not significantly reduce bias magnitude. Crucially, stronger implicit associations significantly predicted discriminatory choices in downstream medical decision tasks (p < 0.001). CONCLUSION: Current safety alignment techniques fail to eliminate implicit biases in large language models within the medical domain. These latent associations translate into biased decision-making, posing risks for health equity. Future development must prioritize representational debiasing over superficial alignment. Furthermore, healthcare professionals must embrace a stance of "AI vigilance": they should critically evaluate algorithmic outputs as fallible "second opinions" rather than objective truths, thereby ensuring that human judgment remains the ultimate safeguard for equitable patient care.
20 May 2026
Read appraisal →JMIR medical education
Experiences and Perceptions of Clinical and Graduate Medical Students Regarding AI in Syria: Cross-Sectional Study
BACKGROUND: Artificial intelligence (AI) tools have revolutionized various aspects of education and health care in recent years. Their influence extends across multiple domains of medical education, from traditional learning to research and foreign language acquisition. OBJECTIVE: This study aims to evaluate the experiences and perceptions of AI tools usage in a low-resource setting and identify the factors influencing their adoption. METHODS: A cross-sectional study was conducted to evaluate the experiences with AI tools and perceptions regarding their future applications in education and health care among medical students in Syria. The sample was equally divided between clinical-year students and graduates. Chi-square tests analyzed differences based on demographics and experience, while Mann-Whitney U tests compared group perceptions of AI's future role. Factors studied included academic year, gender, German language learning, computer access, and research experience. RESULTS: Among 400 participants, AI tools were widely used for study preparation (228/400, 57% of participants), assignments (160/400, 40% of participants), and research. Clinical students used AI more than graduates for examination preparation (P<.001), creating cases (P=.03), and writing tasks (P<.001). Males used AI more for research (P=.004) or anatomy (P=.02); German learners relied on AI for language tasks. Despite 76% (304/400) of students believing AI would enhance residency training and 71.8% (287/400) of students supporting institutional policies, only 25.5% (102/400) of students expected career benefits. Ethical concerns were higher among females and researchers. CONCLUSIONS: This study highlights the increasing reliance on AI tools among medical students and graduates for academic and clinical purposes. The highest usage was reported in study preparation, writing tasks, and clinical simulations. Significant differences in AI usage were observed based on academic level, gender, access to technology, and research experience. While perceptions were largely positive, concerns remained around ethical use, potential job displacement, and diminished human interaction in medicine. These findings underscore the importance of developing institutional policies to guide the ethical and effective integration of AI in medical education.
20 May 2026
Read appraisal →Journal of medical Internet research
Understanding Remission of Long-Term Conditions Through Electronic Health Records: Scoping Review
BACKGROUND: Multiple long-term conditions (MLTCs) require complex and prolonged treatment regimens. Remission in long-term conditions (LTCs) is important for understanding disease progression and evaluating treatment effectiveness. Electronic health records (EHRs) are increasingly used to monitor clinical outcomes, but how remission is defined within EHRs remains unclear. OBJECTIVE: This study aimed to summarize and collate the previous literature on how remission of LTCs has been defined in EHRs. METHODS: Systematic electronic searches were performed on OVID MEDLINE, Embase, CINAHL EBSCO, the Cochrane Library, and the Bielefeld Academic Search Engine for eligible studies published from inception to November 27, 2025. Quantitative studies, published in any language, on adult populations, and using EHRs to assess remission of LTCs, were eligible for inclusion. Studies that did not clearly define remission and studies on cancer remission were excluded. Data were extracted from each eligible study using a structured table. Risk of bias was not assessed, in line with scoping review methodology. A narrative approach was taken to summarize and present data from the included studies. The number and characteristics of studies were described, both overall and by condition. Findings were discussed with clinicians and data experts to ensure applicability in clinical practice. RESULTS: Ninety-one studies were included. Sample sizes ranged from 12 to 72.9 million adults. Studies were conducted in 18 countries, with the majority being from the United States. The majority of included studies used a cohort study design. Studies assessed how remission was defined in 12 LTCs, including inflammatory bowel disease (41/91, 45.1%), type 2 diabetes (n=15, 16.5%), depression (n=15, 16.5%), alcohol or drug misuse (n=8, 8.8%), asthma (n=3, 3.3%), multiple sclerosis (n=3, 3.3%), epilepsy (n=1, 1.1%), anemia (n=1, 1.1%), chronic kidney disease (n=1, 1.1%), autoimmune pancreatitis (n=1, 1.1%), hypertension (n=1, 1.1%), heart failure (n=1, 1.1%), and MLTC (n=1, 1.1%). Remission was typically defined using a combination of clinical codes (n=7, 7.7%), validated rating scales (n=56, 61.5%), biochemical markers (n=29, 31.9%), absence of symptoms (n=10, 11%), absence of condition-specific events (eg, hospital admissions; n=4, 4.4%), and cessation of pharmacological treatments (n=26, 28.6%). There was substantial variation in the criteria and duration of follow-up used to define remission across studies. CONCLUSIONS: This review demonstrates that remission of LTCs can be identified and operationalized within EHRs, although remission criteria varied across studies. The review extends the literature on remission in EHRs by combining evidence synthesis and consultation with clinical and data experts to propose standardized comprehensive definitions to reliably define and implement remission of multiple LTCs in EHR-based research. This will allow cross-study comparisons and present an opportunity to advance understanding of disease trajectories and improve evaluation and monitoring of patient outcomes. Further research may apply, compare, and evaluate standardized definitions across different data sources to assess generalizability and further improve our understanding of remission of LTCs.
20 May 2026
Read appraisal →JMIR infodemiology
Performance of Large Language Models in the Cognitive Analysis of Misinformation: Evaluation Study
BACKGROUND: Public discourse is significantly impacted by the rapid spread of misinformation on social media platforms. Human moderators, while capable of performing well, face many challenges due to scalability. While large language models (LLMs) show great potential across various language tasks, their capacity for cognitive and contextual analysis, in detecting and interpreting misinformation, remains less evaluated. OBJECTIVE: This study evaluates the effectiveness of LLMs in detecting and interpreting misinformation compared to human annotators, focusing on tasks requiring cognitive analysis and complex judgment. Additionally, we analyze the influence of different prompt engineering strategies on model performance and discuss ethical considerations for using LLMs in content moderation systems. METHODS: We evaluated 4 OpenAI models against a panel of human annotators using a subset of posts from the MuMiN dataset. Each model and human annotator responded to structured questions on misinformation, following an established cognitive framework. Both human annotators and LLMs also provided scores indicating how confident they were in their responses. Various prompting strategies were used in this research, including: 0-shot, few-shot, and chain-of-thought, with performance evaluated through precision, recall, F1-score, and accuracy. We used statistical tests, including the McNemar test, to quantitatively assess differences between LLM and human ratings of misinformation. RESULTS: GPT-4 Turbo with chain-of-thought prompting achieved the highest performance of all LLMs for detecting misinformation, with an accuracy of 67.2% and an F1-score of 78.3%, but was outperformed by human annotators, who achieved 70.1% accuracy and an F1-score of 81%. LLMs performed well in tasks involving logical reasoning and straightforward misinformation detection, but struggled with complex judgments, including detecting sarcasm, understanding misinformation, and analyzing user intent. LLM confidence scores positively correlated with accuracy in simpler tasks (r=0.72, P<.01) but were less reliable in subjective and complex contextual evaluations. CONCLUSIONS: LLMs show significant potential for automating misinformation detection. Their limitations in understanding and interpreting these posts highlight the current necessity of human oversight. A hybrid framework combining LLMs for preliminary screening with human moderators for more complex evaluation presents a promising future direction. Future research could prioritize the fine-tuning of LLMs using datasets that emphasize cognitive and emotional linguistic features, alongside the development of advanced prompting techniques.
19 May 2026
Read appraisal →Acta oncologica (Stockholm, Sweden)
KaSARi: a national framework for standardized, automated, and predictive radiotherapy in Finland.
BACKGROUND AND PURPOSE: National radiotherapy (RT) data infrastructures are emerging to support treatment quality assurance and outcome research. However, prospective nationwide integration of full DICOM-RT data with toxicity and patient-reported outcome measures (PROMs) remains uncommon. KaSARi (Key advances in Standardizing, Automating, and predicting Risks in RT) was established to create national research infrastructure for prospective RT data collection and integration in Finland. MATERIALS AND METHODS: KaSARi is a prospective, consent-based multicenter cohort infrastructure involving Finnish university and central hospitals. All adult patients receiving RT are eligible. DICOM-RT datasets from treatment planning and oncology information systems are linked to clinician-graded toxicity, PROMs and relevant clinical variables. The infrastructure builds on previously validated multi-institutional DICOM workflows and follows FAIR (Findable, Accessible, Interoperable, Reusable) data principles. Predefined work packages support data harmonization, scalable data processing and enable downstream applications such as AI-based segmentation, dose prediction, and outcome modeling. RESULTS: Patient recruitment started in May 2025. By February 2026, 221 patients had been enrolled at the coordinating center, initially including breast cancer patients and subsequently expanding to all RT indications. Recruitment at a second center began in November 2025, with 37 patients enrolled. The infrastructure is projected to include approximately 29,000 patients over 15 years. Integration of dosimetric, clinical and PROM data is progressing through a phased national implementation as additional centers join the network. INTERPRETATION: KaSARi represents a national infrastructure prospectively integrating full DICOM-RT datasets with clinician-graded toxicity and PROMs across all RT indications within a unified research protocol. The study provides a foundation for nationwide outcome modeling, harmonization of RT practices and development and validation of data-driven AI-based methods.
19 May 2026
Read appraisal →Journal of global health
Unveiling the gap in heart failure: a Brazilian Unified Health System study.
BACKGROUND: Heart failure (HF) decompensation is the leading cause of hospitalisations in developed countries and the third most common cause in Brazil. Underdiagnosis or misdiagnosis thereof remains a critical challenge and with significant implications. At the patient level, it delays appropriate treatment and disease management. At the health system level, it leads to inefficient population health strategies, distorted epidemiological estimates, and suboptimal resource allocation. Addressing this hidden HF population is therefore essential for improving disease surveillance, optimising care pathways, and supporting more effective and equitable health system planning in Brazil. In this sense, we aimed to quantify underreporting of HF in individuals in Brazil and to estimate their mortality trends. METHODS: We identified potential HF patients using a guideline-based flowchart that combines ICD-10 codes (e.g. I21, I25, I42) and HF-related procedures (e.g. echocardiography, B-type Natriuretic Peptide (BNP) testing). Then, we developed a machine learning model for early prediction of HF using the Brazilian Department of Informatics of the Unified Health System (Departamento de Informática do Sistema Único de Saúde) mortality information system and ambulatory information system records, 2018-2022. A CatBoost algorithm was trained on a balanced cohort of patients (10 000 with HF and 9 995 without HF), restricting predictors to ≥12 months before the first I50 code to prevent leakage. Key predictors included age, chronic kidney disease, echocardiography, and lipoprotein disorders. We validated model performance in independent cohorts, including a low-prevalence cohort (~2% HF). Finally, we derived underreporting estimates via deterministic sensitivity analysis across scenarios from 0-100%. RESULTS: The proxy identified approximately 54 000 potential HF cases/y that were unreported in ambulatory data, with deterministic analysis suggesting 12-41% underestimation. Mortality analysis showed that around 200 000 deaths could be linked to unreported HF. The CatBoost model achieved an area under the curve (AUC) of 0.91 (balanced cohort; accuracy = 0.82, recall = 0.81, F1 = 0.82) and maintained strong discrimination in low-prevalence settings (AUC = 0.84, sensitivity = 0.82), with excellent calibration (Brier = 0.124-0.136; ECE = 0.01-0.03). CONCLUSIONS: We found that HF underreporting in Brazil is substantial and carries significant mortality implications. Our ML model demonstrates high accuracy in early risk stratification using routine administrative data, aligning with clinical pathways. Implementing it as a screening tool could optimise resource allocation, but ethical considerations around false positives warrant careful deployment. Future work should focus on clinical validation and cost-effectiveness analysis within the Brazilian Unified Health System. Furthermore, our findings highlight significant implications for public health and clinical HF management, emphasising the necessity of strategies that promote early detection of HF and precise case recording. Addressing underestimation is crucial to optimise healthcare resources and improve patient outcomes. The results underscore the importance of accurate diagnosis and comprehensive management approaches for better HF case tracking. Further research should explore the public health impact of these underestimations in Brazil, particularly regarding its health system's financial resources.
18 May 2026
Read appraisal →World journal of urology
Comparative study on the efficacy of large language models (LLMs) in health management for patients with urological and andrological diseases: a multidimensional quality evaluation and applicability analysis
PURPOSE: This study aims to evaluate the efficacy of large language models (LLMs) in health management for urological and andrological conditions by comparing two Chinese LLMs: DeepSeek and Kimi. METHODS: The responses of DeepSeek and Kimi to 30 questions on six urological and andrological diseases were quantitatively assessed using quality assessment tools, expert assessments, and readability metrics. RESULTS: In quality assessments, DeepSeek outperformed Kimi in only two instances: responses to "What are the surgical options for kidney stones?" (PEMAT-AI: P = 0.012; DISCERN-AI: P = 0.024) and "What are the treatment modalities for bladder cancer?" (PEMAT-AI and DISCERN-AI: P < 0.001). Expert assessments revealed that DeepSeek demonstrated superior accuracy and safety compared to Kimi only in addressing "Dietary considerations for kidney stone patients" (accuracy: P < 0.001; safety: P = 0.02). No statistically significant differences were found in quality assessments or expert assessments between the models for disease-specific categories. However, Kimi outperformed DeepSeek in readability evaluations. CONCLUSION: Both DeepSeek and Kimi exhibit substantial efficacy and clinical applicability in managing urological and andrological diseases. To enhance their practical use, Kimi requires improvements in accuracy and safety, while DeepSeek should focus on improving readability.
18 May 2026
Read appraisal →Scandinavian journal of primary health care
Shoulder-to-shoulder: healthcare professionals' experiences of learning during electronic health record implementation in municipal primary care.
BACKGROUND: Formal training is commonly used in electronic health record (EHR) implementation, but often fails to address healthcare professionals' ongoing, context-specific learning needs. There is limited knowledge of how learning unfolds in practice when such training is experienced as insufficient, particularly in municipal healthcare services. METHODS: This qualitative study examined learning processes during the implementation of a new EHR system in a municipal healthcare institution in Central Norway. Data was generated through interviews and observation, and was analyzed thematically. RESULTS: Participants experienced formal training as poorly aligned with their digital competencies and everyday clinical work. Limited time and staffing further constrained participation in formal training and follow-up. Alongside formal training, employees engaged in peer-based learning practices embedded in everyday work, described as 'shoulder-to-shoulder' learning. These practices became particularly important due to limited time and staffing, enabling rapid problem-solving and mutual support, while also reinforcing shared responsibility for learning. Rather than merely compensating for inadequate training, shoulder-to-shoulder learning emerged as an important adaptive practice that supported problem-solving and collaboration during EHR implementation. CONCLUSION: Learning to use EHR systems in municipal primary care should be understood as a socially embedded and ongoing process. While formal training may provide initial orientation, implementation strategies should also recognize and support peer-based learning practices that support ongoing learning and collaboration during periods of organizational change.
18 May 2026
Read appraisal →Clinica chimica acta; international journal of clinical chemistry
A laboratory-driven clinical decision support system algorithm for personalized and automated cardiovascular risk stratification in primary care patients.
OBJECTIVE: To develop and evaluate an automated clinical decision support system (CDSS) capable of computing both categorical and exact-percentage cardiovascular risk (CVR) in routine clinical practice. METHODS: We developed and implemented an automated CDSS (AlinIQ®) for CVR assessment within the Clinical Laboratory Department of Hospital Universitari i Politècnic La Fe (València, Spain), applied to patients referred from Primary Care. The analytical profile-requested either via automatic trigger or proactive clinician order-included total cholesterol, HDL-C, LDL-C, non-HDL-C, triglycerides, lipoprotein(a) [Lp(a)], serum creatinine, and estimated glomerular filtration rate. Additional required inputs were smoking status, systolic blood pressure, country of origin, and age at diabetes onset. The CDSS automatically computed SCORE2, SCORE2-OP, SCORE2-Diabetes, and SCORE2 Asia-Pacific, generating both categorical strata and exact-percentage CVR. Lp(a)-adjusted CVR was derived using coefficients from the Spanish Atherosclerosis Society. Following the ESC 2025 Focused Update, patients meeting predefined clinical criteria for moderate, high, or very high CVR were directly assigned to the corresponding category without SCORE calculation. The system also incorporated modifying risk factors, generating standardized interpretive comments and personalized lipid targets. RESULTS: Over one month, 2289 screenings were requested; 189 (8.3%) were excluded. A total of 171 patients (7.5%) were classified as high risk based on score-based calculation (164 [95.9%] triggered automatically), and 29 (1.6%) as very high risk (23 [79.3%] triggered automatically). 119 individuals (6.8%) were reclassified to a higher risk category after Lp(a)-based adjustment. CONCLUSION: This CDSS provides a scalable and reproducible framework for laboratory-driven cardiovascular prevention.
17 May 2026
Read appraisal →Computer methods and programs in biomedicine
Prompt-to-policy: Leveraging large language models to guide deep reinforcement learning in public health emergencies
Rapid and effective decision-making is critical in public health emergencies, where resource allocation must balance multiple objectives under uncertain conditions. Traditional optimization methods often struggle with computational complexity and real-time, heterogeneous data. To address these challenges, we propose a hybrid intelligent agent combining an enhanced Double Deep Q-Network (D2QN-JDA) with large language models (LLMs). The D2QN-JDA improves learning stability and adaptability through joint state-action inputs, dynamic exploration rate, and adaptive reward normalization. The LLM component uses Retrieval-Augmented Generation (RAG) to integrate structured and unstructured data for real-time decision support. Experiments based on data from the Hong Kong COVID-19 outbreak show that the D2QN-JDA outperforms dynamic programming, greedy algorithms, genetic algorithms, and Q-learning, achieving reductions in cost. The LLM component also outperforms manual and regex methods in both single- and multi-point data extraction, enhancing accuracy, recall, F1 score, cost, and time. Our framework effectively addresses complex, multi-objective resource allocation in public health crises.
17 May 2026
Read appraisal →PloS one
Design and evaluation of a resilient IBN architecture: Integrating post-quantum cryptography with adaptive threat detection using machine learning
As the domain of network security keeps on evolving rapidly, especially in sensitive areas such as healthcare systems, the demand for reliable device verification, controlling access, and spotting threats is growing sharply. This paper presents the design, implementation, and systematic evaluation of an improved Intent-Based Networking (IBN) system that integrates post-quantum cryptography, certificate-based identity management, and machine learning-based anomaly detection within a unified framework. The system incorporates SPHINCS+ post-quantum digital signatures for quantum-resilient authentication, X.509 certificate lifecycle management for establishing device trust, and hardware-aware cryptographic operations to maintain efficiency. It further enforces fine-grained access policies using Role-Based Access Control (RBAC) augmented with Multi-Factor Authentication (MFA), ensuring strong access governance across network segments. For early threat detection, machine learning models such as Isolation Forest and MiniBatch KMeans are employed to learn communication patterns and detect anomalous device behavior. Additionally, event logs are maintained using asynchronous, hash-chained logging mechanisms inspired by blockchain principles, ensuring auditability and data integrity. To address evaluation transparency and rigor, the framework is assessed using a controlled prototype testbed with explicitly defined traffic features and reproducible experimental settings. The evaluation considers cryptographic correctness, access control performance, anomaly detection capability, and scalability under increasing workloads. Experimental results demonstrate 100% success in post-quantum signature generation and verification, effective anomaly detection with no observed false negatives in the evaluated scenarios, and stable log-processing throughput as the number of events grows. Importantly, this work does not claim novelty in individual components, but contributes through the system-level integration and empirical evaluation of a quantum-safe, ML-assisted IBN security architecture. The findings highlight key trade-offs between security enforcement and usability, while also identifying limitations such as certificate expiry handling gaps, conservative policy behavior, and lack of large-scale statistical validation. These observations establish a reproducible baseline and motivate future work toward statistically rigorous validation, real-world deployment, and adaptive policy optimization.
17 May 2026
Read appraisal →JMIR medical informatics
Implementation and Evaluation of an Alternative Electronic Health Record Tool for Ordering Blood Products in Pediatric Oncology and Stem Cell Transplantation: Mixed Methods Analysis
BACKGROUND: Repeated blood product ordering is associated with order entry errors and potential patient harm. Traditional electronic health record order sets require repeated re-entry for recurrent transfusions, creating inefficiencies and opportunities for error, and contributing to physician burnout. Historically, we have used order sets to order blood products, which must be re-entered each time a transfusion is needed. Reusable transfusion therapy plans may address these challenges by standardizing and streamlining transfusion workflows. We conducted a pre-post study at a single pediatric academic center, evaluating the implementation of reusable transfusion therapy plans for packed red blood cells and platelets in oncology patients and those undergoing hematopoietic stem cell transplantation. OBJECTIVE: The primary outcome was to evaluate the proportion of transfusions originating from the transfusion therapy plans during the postimplementation period. Secondary outcomes included evaluating (1) the proportion of eligible patients with applied transfusion therapy plans, (2) changes in transfusion efficiency (time from laboratory result to transfusion release and administration, premedication timing, and estimated overnight pages), and (3) the impact on safety (guideline-concordant dosing, irradiated product ordering, and transfusion thresholds). We also assessed health care practitioner experience using an adaptation of the technology acceptance model survey. METHODS: The prestudy period consisted of the 1-year preimplementation, and the postperiod consisted of the 1-year post implementation. We used our institution's enterprise data warehouse (SickKids Enterprise-Wide Data in Azure Repository) to obtain demographic and transfusion details for all eligible patients. The adapted technology acceptance model survey was administered to eligible oncology clinicians. RESULTS: The preimplementation cohort had 558 unique patients who received a total of 2678 transfusions. The postimplementation cohort had 521 unique patients who received 2777 transfusions. During the postimplementation period, 59% of transfusion orders originated from a therapy plan, increasing to 71% in the final month. Compared with order sets, therapy plan-derived transfusions were released and administered significantly faster following laboratory results (P<.001). Guideline-concordant transfusion volumes increased significantly postimplementation for both packed red blood cells and platelets (P<.001), as did the ordering of irradiated blood products (P<.001). No differences were observed in pretransfusion hemoglobin or platelet thresholds between study periods. Use of therapy plans was associated with an average avoidance of 4 overnight blood product entries per night. Survey responses from nurses and providers demonstrated high perceived usefulness and ease of use, with 95% endorsing continued use. CONCLUSIONS: Reusable transfusion therapy plans improved efficiency, standardized safe ordering practices, and were highly acceptable to clinicians. This longitudinal, noninterruptive electronic health record intervention represents a scalable approach to supporting high-risk transfusion workflows in pediatric oncology.
17 May 2026
Read appraisal →Philosophical transactions. Series A, Mathematical, physical, and engineering sciences
Revisiting Rogers' Paradox in the context of human-AI interaction
People learn about the world in many ways: from conducting experiments to copying others' behaviour. The choices we make about how to learn can impact the collective understanding of a whole population, were others to learn from us. Alan Rogers developed simulations to study these phenomena-where agents could individually or socially learn amidst a dynamic, uncertain world-and uncovered a surprising result: the availability of cheap social learning yielded no benefit to population fitness over individual learning. Rogers' Paradox spawned decades of work to understand factors that favour social learning and better model human cultural development. But what happens when humans can learn from artificial intelligence (AI) systems that are themselves learning from us? We revisit Rogers' Paradox in the context of human-AI interaction and extend the simulations towards a simplified network of humans and AIs learning together about an uncertain world. We examine the impact of several learning strategies on the equilibrium of a society's 'collective world model' and assess levers available to stakeholders in human-AI interactions to change network dynamics. We then model negative feedback loops that may arise from humans learning socially from AI, and consider other open questions that could be explored in our simulation framework. This article is part of the theme issue 'World models in natural and artificial intelligence'.
17 May 2026
Read appraisal →Journal of medical Internet research
Digital Patient Decision Aid for Antiobesity Medications: Mixed Methods Study of Human-Centered Design and Usability Evaluation
BACKGROUND: The global burden of obesity continues to rise, highlighting the need for patient-centered approaches to weight management. Shared decision-making is particularly important in the selection of antiobesity medications (AOMs), as treatment options differ in mechanism, effectiveness, side effects, routes of administration, and cost. Despite this preference-sensitive context, only a few patient decision aids (PDAs) have been culturally and clinically adapted for use in Asian populations. OBJECTIVE: This study aims to design, develop, and evaluate a digital PDA, OptiWeight, to support shared decision-making for AOM selection, incorporating perspectives from health care professionals and patients. METHODS: This mixed methods, multicenter study, conducted between August 2022 and November 2025, applied a 4-stage human-centered design process. An evidence-informed prototype was developed based on clinical guidelines, followed by 2 rounds of usability testing using think-aloud protocols to assess navigation structures, perceived usability (System Usability Scale [SUS]), and cognitive workload (NASA Task Load Index [NASA-TLX]). Semistructured interviews with health care professionals specializing in weight management, guided by the Consolidated Framework for Implementation Research, informed clinical implementation and workflow integration. Finally, patients with overweight or obesity evaluated usability, cognitive workload, and overall user experience in outpatient settings. Qualitative data were analyzed using content analysis, and 1-way analysis of variance examined changes in usability and workload across stages. RESULTS: A total of 174 individuals were included across all study stages (usability testing among adults: n=78; health care professional interviews: n=18; and clinical evaluation among patients: n=78). Iterative usability testing comparing system- and user-controlled navigation structures revealed complementary strengths and limitations, leading to the adoption of a hybrid navigation structure supporting both sequential guidance and flexible comparison. Additional design requirements included the use of icon arrays to enhance risk comprehension and localization features such as treatment cost displays and clarification of socially impactful side effects. Perceived usability increased from initial testing to clinical evaluation (SUS: 60.53-73.65, P<.001), meeting good usability thresholds, while cognitive workload decreased (NASA-TLX: 40.35-16.69, P<.001). CONCLUSIONS: Through a systematic human-centered design process integrating health care professional and patient perspectives, OptiWeight addresses the lack of culturally adapted PDAs for AOM decision-making in Mandarin-speaking populations while capturing user needs-particularly regarding navigation flexibility and risk visualization. The final tool demonstrated good usability and feasibility, and workflow considerations suggest potential for integration into routine weight-management care. Further research is needed to evaluate its impact on decision quality and real-world implementation outcomes.
17 May 2026
Read appraisal →JMIR medical informatics
A Multiassessment and Multiprofessional Agents Approach for Medical Chatbot Risk Estimation: Development and Evaluation Study
BACKGROUND: Assessing chatbot responses across 3 domains-medical, ethical, and legal-is essential to ensuring the safe use of artificial intelligence in health care. Although advancements in the use of large language models (LLMs) show significant improvements in evaluating question-answer datasets, such as multiple-choice medical exams, existing systems use general LLMs without incorporating specialized domain knowledge. They rely on standardized instructions without integrating real-world information, and ensemble methods such as majority voting fail to resolve disagreements among agents, resulting in misclassification and challenges in risk assessment. OBJECTIVE: This study aims to design, develop, and evaluate a synergistic approach for assessing risks associated with chatbot responses using multiassessment (MA) and multiprofessional agents (MPAs). METHODS: We designed and developed an approach consisting of MA and MPA, specifically initial assessment (MA1), which internalizes 3 roles and provides an initial risk estimation, and final assessment (MA3), which aims to reach a final consensus based on the previous assessments (MA1 and MA2), with each using 1 LLM. The verification assessment (MA2) incorporates an MPA or role-based LLM specialized agents for each risk domain (medical, ethical, and legal). We evaluated the proposed approach using the MedNLP-CHAT (Medical Natural Language Processing for AI Chat) corpus (N=226; 100 train, 126 test), covering baseline, enhanced prompt, embedding-based search, and retrieval-augmented generation (RAG). Primary metrics included macro F1-score and joint accuracy to evaluate system performance, along with CI and paired macro F1-score difference (Δ) as supporting metrics to assess the approach's effectiveness. RESULTS: The MA-MPA framework integrated with RAG achieved the highest average macro F1-score of 0.800 across risk domains and a joint accuracy of 76 (60.3%) correct predictions across all risk domains out of 126 question-answer pairs, with notable improvements over the best reported Eighteenth NII Testbeds and Community for Information Access Research Project (NTCIR-18) MedNLP-CHAT systems in the ethical (+0.252) and legal (+0.096) risk domains, while the medical domain showed a modest increase of +0.070. The MA approach contributed the largest gains, particularly from MA1 to MA2, with paired macro F1-score gains ranging from +0.176 to +0.214 across systems. The MPA approach performed better when integrated with MA and external knowledge, with paired bootstrap estimates showing a gain of +0.037 (95% CI 0.003-0.074) over baseline; however, joint accuracy gains were not evident (95% CI -2.9% to 7.7%), and gains relative to the enhanced prompt were small. Notably, MA alone achieved higher joint accuracy than RAG (62.7% vs 60.3%), indicating a metric-specific trade-off rather than consistent superiority across all metrics. CONCLUSIONS: The MA-MPA approach shows potential for improving risk estimation in chatbot responses. The results suggest that the framework is particularly useful for enhancing balanced overall performance, especially when combined with external knowledge, although the medical risk domain remains challenging. Furthermore, more specialized LLMs may further improve contextually grounded risk estimation.
17 May 2026
Read appraisal →JMIR mental health
Factors Influencing the Initiation and Continued Engagement of Digital Mental Health Tools Among Adults: Theory of Planned Behavior-Informed Systematic Review
BACKGROUND: Digital mental health tools (DMHTs) offer scalable support, but engagement varies. Understanding the shapes of initiation and ongoing use is essential for effective design and implementation. OBJECTIVE: This study aims to synthesize determinants of adults' initiation and engagement with DMHTs, organized through two lenses: (1) psychological factors aligned with the theory of planned behavior (TPB) and (2) design and access features. METHODS: A systematic search of 9 databases (June 2025) identified qualitative and mixed methods primary studies reporting end-users' experiences with DMHTs. Studies were screened and reported in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Quality appraisal used quality assessment with diverse studies (QuADS). Data were synthesized using a framework-guided thematic approach, mapping findings to TPB constructs and complementary design and access domains. RESULTS: A total of 22 studies met inclusion criteria. Findings clustered into 2 interdependent domains. TPB constructs explained how beliefs, social expectations, and perceived control shaped decisions to start and persist with DMHTs. Design and access features frequently acted through these same pathways, especially by altering perceived behavioral control (PBC), with cost, connectivity, device constraints, and time flexibility affecting feasibility, with content design and privacy shaping perceived value and trust. Perceived fit (goals, cultural or linguistic relevance, and routine alignment) consistently influenced both initiation and continuation. Several features operated bidirectionally; depending on context, the same feature could facilitate or hinder engagement. CONCLUSIONS: Engagement with DMHTs is jointly determined by users' beliefs and the design and access conditions within which tools are offered. Implementation should pursue a dual strategy, strengthening willingness to seek support (addressing attitudes, norms, and perceived control) while engineering low-effort, trustworthy, and context-appropriate experiences. Priorities include equity-focused policies (data costs, devices, and connectivity), transparent data practices, co-design with diverse communities, and consistent, theory-informed outcome measures.
17 May 2026
Read appraisal →Injury prevention : journal of the International Society for Child and Adolescent Injury Prevention
Large language models in public health: opportunity or threat? The case of button battery injuries.
BACKGROUND: Button battery (BB) injuries in children represent a severe and growing public health burden. The literature on the topic is extensive; however, there is a notable lack of structured public health initiatives addressing the problem. The present study aimed to test the feasibility of using large language models (LLMs) to draft recommendations for preventing and managing BB ingestion in children. METHODS: A set of questions was generated and submitted to ChatGPT-4o and ChatGPT-o1-preview. Questions were based on statements and websites of scientific societies and not-for-profit organisations and were developed to produce comprehensive recommendations that provided information on BB risks, primary and secondary prevention, clinical management and follow-up, and general public health initiatives. Two independent reviewers rated the accuracy and readability of the questions submitted to the LLMs. The accuracy was rated using a four-level scale, while the readability was assessed using two established readability tools, the Flesch Reading Ease (FRE) and the Flesch-Kincaid Grade Level (FKGL). RESULTS: None of the answers provided by the LLMs were rated as completely incorrect or partially incorrect. ChatGPT-o1-preview outperformed ChatGPT-4o in accuracy, with eight answers rated as accurate and complete. Both models showed similar readability levels, with high FKGL and FRE scores indicating college-level comprehension. DISCUSSION: LLM demonstrated a strong performance in this study, with no responses rated as incorrect or partially incorrect, showing its great potential and feasibility for use in public health. CONCLUSIONS: The present findings suggested the potential feasibility of LLMs in public health for preventing paediatric injuries.
17 May 2026
Read appraisal →European journal of pediatrics
Leveraging artificial intelligence to predict non-adherence to pediatric immunization schedules: a systematic review
UNLABELLED: Immunization is one of the most effective interventions to prevent infectious diseases. Identifying individuals at risk of non-adherence to immunization schedules could enable early interventions to improve coverage. With the increasing availability of large datasets with sociodemographic, medical, and geographic data, artificial intelligence (AI) offers new possibilities for identifying predictive patterns. We performed a systematic literature review of original studies that used AI to predict immunization non-adherence in children. Searches were conducted in PubMed, Embase, and Web of Science and no restrictions were applied regarding publication year, language, or study design. Study characteristics, dataset size, AI methods, performance metrics, and validation approaches were extracted. Risk of bias was assessed using PROBAST + AI. A total of 843 records were identified; after screening, 12 studies were included. Nine of 12 studies were conducted in low- and middle-income countries, used cross-sectional survey data, and analyzed vaccines from routine immunization schedules. Model performance was generally high; 9/12 studies reported an accuracy or an area under the receiver operating characteristic curve exceeding 80%, but no study conducted external validation. Predictors clustered into seven categories: maternal, paternal, family, child, geographic, wealth, and others. Among maternal and child-related factors, prenatal care and previous immunization history had a high predictive weight. CONCLUSION: AI demonstrates a promising potential for identifying individuals at risk of vaccine non-adherence. However, the lack of external validation raises concerns about model overfitting and limits real-world applicability. Demographic health surveys were routinely repeated across countries and time periods and could therefore represent an ideal resource for external validation. WHAT IS KNOWN: • Geographic, sociodemographic, and medical factors contribute to non-adherence to the immunization schedule. • Identifying these individuals can guide early interventions to improve immunization coverage. WHAT IS NEW: • Artificial intelligence can reveal unknown influential factors; prenatal care and delivery setting emerged as key factors in this systematic review.
17 May 2026
Read appraisal →JMIR human factors
A Chronic Pain Self-Management Mobile App (Dolodoc): Cross-Sectional Acceptability Study
BACKGROUND: Approximately 19% of adults in Europe are affected by chronic pain, which reduces the quality of life. Pain-management mobile health (mHealth) apps offer a promising solution for self-management, but user engagement and adherence can limit their clinical impact. User experience design and research play a crucial role in optimizing usability and long-term adoption of digital health interventions. OBJECTIVE: This study aims to evaluate the user experience of Dolodoc, a mobile app for chronic pain self-management, using a mixed methods approach that assesses acceptability through a content quality survey and examines use by analyzing overall use patterns. METHODS: A cross-sectional acceptability study of the main content of Dolodoc was conducted among patients with chronic pain recruited from the Geneva University Hospitals pain center and through snowball sampling. Participants rated 84 evidence-based self-management strategies by using a 5-point Likert scale based on 5 acceptability criteria: understandability, motivation, feasibility, relevance, and alignment with the related quality-of-life dimension. To reduce participant fatigue and avoid random responses, the 84 strategies were randomly divided across survey versions. Use was assessed through metrics collected over 6 months with Piwik PRO Analytics to observe real-world use behaviors among Dolodoc users. RESULTS: In the acceptability study, 33 participants rated the self-management strategies positively across all dimensions. On a scale from -2 to 2, the strategies were well understood (mean 1.47, SD 0.76), motivational (mean 1.12, SD 0.96), feasible (mean 1.01, SD 1.05), relevant (mean 0.99, SD 1.09), and aligned with the dimensions (mean 1.33, SD 0.89). The use study demonstrated that 60% (486/802) of the patients used Dolodoc only once, indicating that long-term adherence remains a challenge. Within Dolodoc, pain tracking, useful links, and medication logging were the most actively used features. CONCLUSIONS: This study highlights the gap between acceptability and long-term adherence to mHealth solutions. Improving personalization and accessibility could increase user engagement and long-term adherence. Future iterations of the app should incorporate tailored interventions and real-time feedback mechanisms. In addition, leveraging a digital navigation follow-up could facilitate user adoption and sustained engagement.
17 May 2026
Read appraisal →JMIR mHealth and uHealth
Engagement of Users in Digital Health Applications: Scoping Review
BACKGROUND: Mobile health (mHealth) uses mobile technology as a tool for prevention and health promotion. Research indicates that user engagement is crucial for effective mHealth interventions and improved health outcomes. However, many studies report low adoption rates, rapid decline after initial use, and a lack of acknowledgment of user implications in achieving outcomes. Thus, conceptualizing participation in mHealth is essential to identify key determinants for engaging users. OBJECTIVE: This scoping review aims to identify and characterize the attributes and definitions of user engagement in mHealth, examine engagement methods, and analyze barriers and facilitators influencing participation. METHODS: Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines, Scopus, Web of Science, and PubMed databases were searched for publications between 2000 and 2025 with a 2-stage selection process. RESULTS: Out of 2489 articles identified, 1416 were screened, and 52 met the inclusion criteria. Half were recently published in the last 5 years (2020-2025). Existing literature focused on digital interventions for specific populations and health topics. Analysis revealed four main perspectives on engagement in mHealth: (1) usage metrics, (2) subjective user experiences, (3) a hybrid approach that combines both, and (4) a goal-oriented perspective (behavior change or health outcomes). CONCLUSIONS: To understand the complexity and multifactorial nature of participation, it is relevant to conceptualize it as a dynamic mechanism enabling users to achieve their objectives. Both quantitative use and subjective user experience should be integrated to reach the optimal intervention dose. Recognizing users' evolving needs, uniqueness, and their socioenvironmental context interdependence, it is essential to involve users in all stages (design, implementation, and iterative evaluation of mHealth). Findings will inform an e-Delphi study to establish consensus on engagement criteria.
16 May 2026
Read appraisal →JMIR formative research
Use of Commercially Available Large Language Models to Generate Information Leaflets on Post-Intensive Care Syndrome: Clinical Utility Assessment
BACKGROUND: Patients and their families without medical knowledge may find professional health care information difficult to understand. The use of large language models (LLMs) to simplify and translate complex medical content holds promise for improving comprehension while reducing the burden on health care providers tasked with delivering explanations. OBJECTIVE: This study aims to evaluate the quality of information leaflets generated using commercially available LLMs. METHODS: Informational texts on post-intensive care syndrome were generated using 6 different LLMs and 4 prompt designs with varying levels of instructional guidance. Clinical practice guideline documents were uploaded and provided to the models as reference context, reflecting a pragmatic clinical scenario without model tuning or advanced retrieval pipelines. In total, 72 texts were generated (6 models × 4 prompts × 3 outputs). After excluding texts shorter than 500 characters (n=16) and those without explicit mention of post-intensive care syndrome (n=3), 53 texts remained. To enable balanced human evaluation across model-prompt combinations, the longest eligible response from each pair was selected (4 prompts × 4 models; n=16). Following independent expert review by 2 medical specialists, 7 texts were excluded, leaving 9 texts for final analysis. Ten individuals, including health care professionals and nonmedical personnel, assessed the texts using a 10-point Likert scale across multiple quality domains. An LLM-based parallel assessment was also conducted, and scores were compared across models and evaluator groups. RESULTS: In the human evaluation of the selected 9 texts, the generated texts achieved an average score of 6.8 or higher across all evaluation criteria, with no potentially harmful content identified. The text generated by LLaMA 3 70B, using a step-by-step approach combined with text-augmented prompting based on clinical guidelines, received the highest overall score, whereas the lowest-rated text was produced using a simple prompt without text augmentation. Although no consistent trends were observed across LLMs or prompt engineering strategies, text-augmented prompting was generally associated with higher evaluation scores. Ratings differed between professional and nonprofessional evaluators. Given the feasibility-driven screening process and the resulting limited sample size, the findings should be interpreted as exploratory and descriptive rather than definitive estimates of overall model performance. CONCLUSIONS: Among the selected texts included in the final human evaluation, informational materials generated using commercially available LLMs were generally rated as acceptable by human evaluators, and none contained harmful content. These findings suggest that LLMs may support the development of patient-facing informational materials under feasibility-constrained conditions, although further validation with larger and more diverse samples is warranted.
16 May 2026
Read appraisal →BMJ open
Transitioning from paper-based to electronic health management information systems in Africa: a scoping review protocol
INTRODUCTION: The rising shift from paper-based to electronic health management information systems (EHMIS) among global health systems has shown promising strides over the past decade. Yet, most African health systems have continued to use paper records with attendant gaps and challenges. Most African governments have now started transitioning from paper to EHMIS but lack adequate guidance to support this shift. There is therefore a need for harmonised regional guidance to ensure that such transitions are carried out systematically and take into account country-specific contexts. OBJECTIVE: The objective of this study protocol is to conduct a scoping review to generate evidence that will inform the development of a comprehensive guide to support countries in the WHO African Region in transitioning from paper-based to EHMIS. METHODS AND ANALYSIS: The review will follow established methodological guidance for scoping reviews as outlined by Arksey and O'Malley and further refined by Levac et al and the Joanna Briggs Institute, with reporting guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews checklist. A search strategy will be developed to systematically identify relevant studies from both published and grey literature sources including PubMed, Cochrane Library, Scopus and African Index Medicus. Other sources will include Google Scholar, Emerald Journal, the WHO Regional Office for Africa Library and websites of WHO, ITU and Ministries of Health. Reference lists of the retrieved published articles will be manually searched to identify additional relevant studies. Descriptive qualitative content analysis will be undertaken using the policy analysis framework and key findings will be summarised and presented using tables, charts and maps. ETHICS AND DISSEMINATION: This study does not involve the collection of primary data; therefore, ethical approval will not be required. On completion of the study, findings will be submitted for publication in a peer-reviewed scientific journal and will also be presented at national, regional and international conferences to support knowledge sharing and policy engagement.
16 May 2026
Read appraisal →JMIR research protocols
A Natural Language Processing Framework for Structuring and Visualizing Clinical Trial Eligibility Criteria at Scale: Protocol for a Quantitative Study
BACKGROUND: Eligibility criteria are essential to clinical trial design, guiding recruitment, and ensuring patient safety and scientific rigor. However, criteria are often lengthy, heterogeneous, and inconsistently formatted, which hinders large-scale interpretation and slows patient-trial matching. Manual review is time-consuming and error-prone. Advances in natural language processing and large language models (LLMs) offer opportunities to standardize and analyze eligibility text at scale. OBJECTIVE: This study aims to develop and evaluate a scalable system that uses LLM-enabled natural language processing and unsupervised learning to identify, normalize, categorize, and visualize clinical trial eligibility criteria, with the goal of improving patient-trial matching and revealing domain-level trends. METHODS: We designed a three-part pipeline: (1) representation of eligibility text using embeddings, followed by clustering to group semantically similar criteria; (2) dual-layer zero-shot LLM summarization for concept normalization, refinement, and deduplication of cluster exemplars; and (3) an interactive, web-based visualization interface to explore criteria distributions and trends by disease domain and over time. The pipeline was applied to 53,872 oncology trials (breast, lung, and gastrointestinal cancer) indexed on ClinicalTrials.gov. Outputs include cluster labels, normalized criterion summaries, and per-domain frequency profiles. Feasibility was assessed via successful end-to-end processing and inspection of face validity for cluster coherence and domain-specific patterns. RESULTS: The system successfully processed all 53,872 trials and generated stable clusters of inclusion and exclusion concepts. The LLM summarization layers produced concise, nonredundant labels that improved the interpretability of clustered criteria. The visualization interface enabled rapid exploration of cross-trial patterns and temporal trends within breast, lung, and gastrointestinal oncology, facilitating identification of common inclusion requirements and potential barriers to enrollment. A public, open-source demonstration instance allows for interactive exploration of these clusters and summaries. Benchmarking through human validation on a random sample of eligibility criteria found the system to be 94% (470/500) accurate, reflecting its ability to consistently categorize criteria correctly in congruence with human judgment. CONCLUSIONS: A combined embeddings-clustering-LLM pipeline can standardize heterogeneous eligibility text and surface domain-level patterns at scale. This framework provides a foundation for accelerating patient-trial matching and informing future trial design. While the current implementation was evaluated on ClinicalTrials.gov oncology trials, the approach is readily generalizable to additional diseases and alternative modeling configurations.
16 May 2026
Read appraisal →JMIR research protocols
Evaluating the Methodological Quality of Artificial Intelligence-Assisted Systematic Reviews: Protocol for a Mixed Methods Meta-Research Study
BACKGROUND: Artificial intelligence (AI), including large language models (LLMs), is increasingly integrated into systematic review (SR) workflows. AI tools may accelerate searching, screening, data extraction, and reporting, but their effects on methodological quality, reporting completeness, transparency, and reproducibility remain uncertain. Existing evaluations largely examine isolated tasks, and inconsistent disclosure of AI use limits reproducibility and oversight. OBJECTIVE: This 4-phase mixed methods meta-research study will (1) compare the methodological quality of AI-assisted versus traditional SRs; (2) refine, finalize, and apply a preliminary AI Transparency and Disclosure Index (AITDI); (3) evaluate reproducibility by comparing outputs across repeated runs of the same AI model, across different AI models, and between AI models and human reviewers at multiple SR stages; and (4) explore knowledge user perspectives on rigor, transparency, and trust in AI-assisted SRs. METHODS: We will conduct a matched cohort analysis of SRs published from 2023 to 2025 in biomedical journals. Each AI-assisted SR will be matched 1:2 with traditional SRs by publication year, clinical domain, review type, and meta-analysis status. Two independent reviewers will apply A Measurement Tool to Assess Systematic Reviews, version 2 (AMSTAR 2; methodological quality), PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 (reporting completeness), and, when applicable, Risk of Bias in SRs (ROBIS; risk-of-bias rigor). A preliminary AITDI will be refined and then applied to all AI-assisted SRs. Reproducibility will be assessed using SR-derived task sets to compare outputs across repeated runs of the same model, across different models, and between AI and human reviewers at key SR stages. Semistructured interviews with authors, editors, clinicians, policymakers, and patient partners will be analyzed using reflexive thematic analysis. RESULTS: As of December 2025, the study has been preregistered on the Open Science Framework (OSF; DOI: 10.17605/OSF.IO/Q5JRW), the search strategy has been finalized, and title/abstract screening has begun. Data extraction is planned for March-May 2026, followed by AITDI refinement and reproducibility testing from May 2026 to October 2026. Qualitative interviews are anticipated from October 2026 to February 2027, with final analyses by April 2027 and dissemination planned for mid-2027. CONCLUSIONS: This study will provide one of the first empirical comparisons of methodological quality, transparency, and reproducibility of AI-assisted versus traditional SRs in the LLM era. Findings will inform expectations for responsible AI integration and support refinement of reporting and methodological best practices, including future development of AI-specific reporting and appraisal extensions (eg, PRISMA-LLM [Preferred Reporting Items for Systematic Reviews and Meta-Analyses-large language model] and AMSTAR-LLM [A Measurement Tool to Assess Systematic Reviews-large language model]).
16 May 2026
Read appraisal →Clinica chimica acta; international journal of clinical chemistry
Artificial intelligence algorithm optimization and application in patient-based real-time quality control (PBRTQC)
Patient-based real-time quality control (PBRTQC) serves as a vital supplement to quality management in clinical laboratories. Its core principle is to monitor the testing process in real time and continuously through patient test data. As artificial intelligence (AI) technology develops rapidly, AI has provided novel pathways for the innovation of PBRTQC algorithms and drives its transition from a traditional statistics-driven model to intelligent monitoring. This review systematically summarizes the progress of AI-driven PBRTQC algorithm optimization. Meanwhile, it provides a detailed account of the clinical applications of the AI-PBRTQC monitoring platform. These applications encompass timely quality control early warning, homogeneous monitoring across multiple settings, precise quality control in complex clinical settings, anomaly traceability and subsequent correction. In addition, this review offers an in-depth analysis of the challenges that arise during the practical implementation of AI-PBRTQC. These include technical limitations, shortage of professional talents, system compatibility barriers, and lagging standardization and regulation. It also explores future development trends and provides valuable references for the intelligent upgrade of PBRTQC.
16 May 2026
Read appraisal →Journal of medical Internet research
Understanding Patient-Reported Offenses in Electronic Health Records: Cross-Sectional Mixed Methods Survey
BACKGROUND: Patients' access to their electronic health record (EHR) supports their participation and satisfaction with care. Despite the benefits, some patients have been upset after reading their EHR. Additionally, health care professionals are concerned that patients, particularly those with mental health conditions, may be offended, and they have expressed a need for further guidelines on how to write EHRs. Experiences among various patient groups are essential to support the relationship between patients and professionals. However, prior studies have often focused on single patient groups or specific clinical contexts, leaving a limited understanding of differences across multiple patient groups. OBJECTIVE: This study aimed to determine whether certain patient groups are more likely to feel offended while reading their EHRs and which information is perceived as offensive and to provide a comparison across multiple patient groups using a mixed methods approach. METHODS: A cross-sectional survey was conducted via the Finnish national patient portal using a web-based patient survey, adopting a mixed methods approach. The survey included multiple-choice and open-ended questions. The total sample comprised 4681 respondents. The survey respondents were placed into 4 patient groups: those who had received care for mental health, cancer, or other conditions and those who had received no care. Associations between the type of care and patients who felt offended were estimated using multivariate binary logistic regression. Inductive content analysis (n=502) was conducted to identify information perceived as offensive in the EHR. RESULTS: The patients who had received mental health care (166/654, 25.4%) or cancer and mental health care (9/39, 23.1%) were more likely to be offended by information in their EHR compared to the other groups (cancer care: 37/375, 9.9%; other conditions care: 383/3316, 11.6%; no care: 22/206, 10.7%; other conditions care: odds ratio 0.37, 95% CI 0.29-0.46; P<.001; model A). Additionally, female patients, those with bad or very bad health conditions, and patients with bachelor's or master's degrees were significantly more likely to feel offended. Errors, the health care professionals' disrespectful language, and perceived unnecessary information were the most frequently mentioned reasons for being offended. Patients with mental health care reported more often that unnecessary information and professionals' opinions and word choices were experienced as offensive compared to other patients. CONCLUSIONS: This study contributes new knowledge by identifying differences across multiple patient groups. Although a minority of patients felt offended by their EHR, health care professionals should consider that some patients, particularly those who have received mental health care or cancer and mental health care, may be offended by specific information or word choices in their EHRs. To address this, health care professionals should receive education on how to write their notes in a neutral tone and avoid potentially offensive topics. Improving the quality of EHRs could strengthen the relationship between patients and professionals.
15 May 2026
Read appraisal →BMJ open
Prediction of inpatient falls and key predictors using machine learning applied to electronic health records: a retrospective cohort study in a tertiary hospital in Türkiye
OBJECTIVE: Traditional fall risk tools are often inaccurate and burdensome. This study aims to develop a predictive model for inpatient falls and identify the most influential variables using machine learning applied to electronic health record data. DESIGN: A retrospective cohort study. SETTING: A large tertiary university hospital in Türkiye. PARTICIPANTS: Adult patients (≥18 years) hospitalised in a university hospital between January 2017 and June 2023. PRIMARY OUTCOME MEASURES: Occurrence of inpatient falls recorded in incident reporting systems. RESULTS: A total of 518 fallers were identified and compared with 3121 non-fallers. Fallers were significantly older (median 68.5 vs 64 years, p<0.001), had longer hospital stays (16 vs 12 days, p<0.001) and higher comorbidity burden (p<0.001). They were also more likely to receive high fall-risk medications (86% vs 76%, p<0.001) and had a higher prevalence of mental disorders (26.3% vs 17.2%, p<0.001). The random forest quantile classifier demonstrated the highest discrimination (area under the curve (AUC) 0.821, 95% CI 0.781 to 0.861), with balanced sensitivity (0.799, 95% CI 0.734 to 0.867) and specificity (0.722, 95% CI 0.691 to 0.755). Pairwise comparisons using DeLong's test showed significantly higher AUC compared with several models (adjusted p=0.001), while differences with eXtreme Gradient Boosting were not significant (p=0.240). Key predictors included comorbidity burden, age, number of fall-risk-increasing medications and laboratory variables such as eosinophil and basophil counts, blood urea nitrogen, sodium and red blood cell count. CONCLUSION: Machine learning models using electronic health data can predict inpatient falls and reveal key risk factors. The random forest quantile classifier offers a promising approach for improving fall risk prediction in imbalanced clinical datasets.
15 May 2026
Read appraisal →BMJ open
AI chatbots for health information seeking among Chinese patients with precancerous ENT lesions: a descriptive qualitative study
OBJECTIVE: To explore how Chinese patients with precancerous ear, nose and throat (ENT) lesions experience using artificial intelligence (AI)-driven chatbots for health-related information seeking, with particular attention to perceived benefits, challenges and influences on information-seeking practices. DESIGN: Descriptive qualitative study. SETTING: Department of Otolaryngology, West China Hospital, Sichuan University, China. PARTICIPANTS: 12 adult patients with clinically diagnosed precancerous ENT lesions who had used AI-driven chatbots at least three times in the previous 30 days to seek health-related information about their condition were purposively recruited. Face-to-face, semistructured interviews were conducted between 11 October 2024 and 10 November 2024. RESULTS: Interviews were analysed using Colaizzi's method. Four themes were identified. First, participants described AI chatbots as an immediate and accessible source of health information, particularly when questions arose outside clinical encounters or during periods of uncertainty. Second, many reported moving away from conventional search engines towards conversational information seeking, valuing direct and synthesised responses over link-based retrieval. Third, participants emphasised that obtaining useful answers depended on learning to ask clear and specific questions, suggesting that effective prompting was an important user skill. Fourth, some participants perceived chatbot interaction as emotionally safer than asking healthcare professionals certain questions, particularly when they felt embarrassed, worried about asking repetitive questions or feared being judged. CONCLUSIONS: Among this sample of Chinese patients with precancerous ENT lesions, AI-driven chatbots were perceived as a convenient and conversational supplementary source of health information. Participants valued their accessibility and interactional ease, but also indicated that their usefulness depended partly on users' ability to formulate effective prompts. Some participants additionally perceived chatbots as a more comfortable channel for asking sensitive or basic questions. The findings suggest that AI chatbots may have a complementary role in patient information support, but further research is needed to evaluate response accuracy, safety and appropriate integration into clinician-led care.
15 May 2026
Read appraisal →Philosophical transactions. Series A, Mathematical, physical, and engineering sciences
Levels of analysis for large language models
Modern artificial intelligence systems, such as large language models, are increasingly powerful but also increasingly hard to understand. Recognizing this problem as analogous to the historical difficulties in understanding the human mind, we argue that methods developed in cognitive science can be useful for understanding large language models. We propose a framework for applying these methods based on the levels of analysis that David Marr proposed for studying information processing systems. By revisiting established cognitive science techniques relevant to each level and illustrating their potential to yield insights into the behaviour and internal organization of large language models, we aim to provide a toolkit for making sense of these new kinds of minds. This article is part of the theme issue 'World models in natural and artificial intelligence'.
15 May 2026
Read appraisal →Systematic reviews
Barriers and facilitators to the implementation and adoption of computerised clinical decision support systems: an overview of reviews
BACKGROUND: The use of computerised clinical decision support systems (CDSSs) holds considerable potential in enhancing healthcare delivery by improving patient safety, practitioner performance, and patient outcomes. However, despite the numerous reported advantages of CDSSs, their adoption remains restricted, thereby compromising the full realisation of their potential. To enable a profound evaluation and successful implementation of these systems, it is imperative to identify the reasons for their limited uptake. The objective of this study is to provide a comprehensive overview of the barriers and facilitators to the implementation and adoption of decision support systems across healthcare settings. METHODS: This study was reported in accordance with the Preferred Reporting Items for Overviews of Reviews (PRIOR) statement. A systematic search was conducted in the databases PubMed, IEEE Xplore, Scopus, and Web of Science from inception to 15 October 2024, targeting at review articles of primary studies focusing on the identification and reporting of barriers and facilitators to the implementation and adoption of CDSSs across healthcare settings. The risk of bias of the included reviews was assessed with the Risk of Bias in Systematic Reviews (ROBIS) tool, and the primary study overlap was calculated using the corrected covered area method. All data on barriers and facilitators to the implementation and adoption of CDSSs were extracted from the included reviews, synthesised through a mapping of identical factors, and ranked according to their frequency of occurrence. RESULTS: Of the 1640 records retrieved through database searches, 30 reviews were included in this overview, which summarised the findings of 721 unique primary studies. A total of 101 distinct factors influencing the implementation and adoption of CDSSs were identified and could be categorised into the overarching categories of human, technology-related, and contextual factors. The following six factors were reported most frequently: usability, usefulness/perceived benefits, organisational readiness, training, trust, and workflow integration. CONCLUSIONS: The findings of this study highlight the diversity of factors that influence a successful implementation of CDSSs and emphasise the need for their comprehensive evaluation that goes beyond the assessment of general performance aspects but takes into account human and contextual factors. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42024507614.
15 May 2026
Read appraisal →BMJ open
Developing and validating an electronic health record-embedded AI model for managing multimorbid hospitalisation risk in patients with chronic RESpiratory disease (AiRES): a study protocol
BACKGROUND: Chronic respiratory diseases (CRDs), such as asthma and chronic obstructive pulmonary disease (COPD), are heterogeneous conditions with a high multimorbidity burden. However, existing risk assessment instruments prioritise physiological measures while overlooking systemic comorbidities. We aim to develop and validate an electronic health record (EHR)-embedded artificial intelligence (AI) model-AiRES (AI in patients with RESpiratory disease)-to predict the 30-day, 90-day and 180-day risks of all-cause and index-disease hospitalisations. This model represents a first step towards a clinical decision support tool for personalised multimorbidity management in patients with CRD. METHOD AND ANALYSIS: Patients aged ≥18 years with a validated case definition of asthma and COPD will be identified from Singapore health administrative data (2012-2020). Candidate predictors will include age, sex, ethnicity, housing type, and comorbidities, measured across multiple care settings as visit frequency, grouped at quarterly intervals in Year 1 and annually for Years 2 and 3 over a 3-year lookback window. We will predict 30-day, 90-day, and 180-day risks of (1) all-cause and (2) asthma/COPD-specific hospital admissions using up to five randomly selected index dates per individual. Three machine learning algorithms-logistic regression (LR) with Lasso regularisation, eXtreme Gradient Boosting, and Categorical Boosting-will be trained using 10-fold cross-validation (CV) with an ensemble feature selection strategy. The optimal model, selected based on performance and feature importance, will be benchmarked against two reference models: a full LR and a Zero-Inflated Negative Binomial regression with hospitalisation history as the sole predictor. Discrimination and calibration will be assessed using internal-external cluster-based and temporal CV. Clinical utility will be evaluated using decision curve analysis. ETHICS AND DISSEMINATION: This study obtained ethics approval from the National University of Singapore (NUS-IRB-2024-849). Results will be published in international peer-reviewed journals.
15 May 2026
Read appraisal →PloS one
Classification of soft decision-making methods via fuzzy parameterized fuzzy soft matrices and their performance-based statistical analysis in machine learning
This study provides a comprehensive evaluation and classification of 35 soft decision-making (SDM) algorithms based on fuzzy parameterized fuzzy soft matrices (fpfs-matrices). Although fpfs-matrices offer a strong mathematical framework for modeling uncertainty, there has been a lack of large-scale comparisons of their derivative SDM methods in machine learning. To address this, we used the Comparison Matrix-Based Fuzzy Parameterized Fuzzy Soft Classifier (FPFS-CMC) to benchmark these 35 algorithms across ten diverse datasets from the UCI Machine Learning Repository. The methods were thoroughly assessed utilizing metrics such as accuracy, precision, recall (sensitivity), specificity, and F1-score, and statistical significance was confirmed using the Friedman and Nemenyi tests. Our results show that SDM methods via fpfs-matrices perform competitively in classification tasks involving uncertainty. Notably, the best algorithms according to F1-scores were A19 (Rank 1), YHX14 (Rank 2), and a three-way tie for Rank 3 among VMH16, AKO18o, and A19/2. By identifying the most effective algorithms and offering a structured decision-support framework, this research provides both a theoretical reference and practical guidance for practitioners selecting SDM methods for complex machine learning challenges.
15 May 2026
Read appraisal →Journal of medical Internet research
Leveraging Text Messaging and Behavior Theory to Improve Colorectal Cancer Screening in Federally Qualified Health Centers: Cohort Study
BACKGROUND: Colorectal cancer (CRC) is the second leading cause of cancer-related deaths in the United States. Despite recommendations for screening to begin at the age of 45 years, significant disparities persist, particularly among medically underserved populations. OBJECTIVE: This study examines the effectiveness of SMS text messaging reminders in improving CRC screening rates across 2 large federally qualified health centers (FQHCs) serving vulnerable populations. METHODS: The study included 4822 adults aged ≥45 years, receiving care at 2 large FQHC networks in Texas and California. Participants were assigned to one of four groups: (1) control (no SMS text messages), (2) single-outreach SMS text overdue message, (3) three-week SMS overdue and reminder text messages, and (4) six-week SMS text messages that were informed by behavior theory. Data were collected from May 2023 to July 2024. The outcome measure was a binary indicator of whether the participant underwent 1 of 3 CRC tests, fecal immunochemical test, colonoscopy, and Cologuard, within 90 days of completing the SMS text messaging reminders. Independent variables included demographic, geographic, clinical, and primary care access variables. Multivariate logistic regression models were used to examine associations between CRC screening completion and the SMS text messaging reminder groups, adjusting for covariates. Adjusted odds ratios (aORs) and 95% CIs were reported. RESULTS: In the combined 3-test model, patients in the single-outreach SMS text message (aOR 1.22, 95% CI 1.00-1.47) and the 3-week SMS text message (aOR 1.27, 95% CI 1.05-1.53) groups had higher odds of completing the screening test compared to those in the control group. Within the fecal immunochemical test-only model, patients in the 3-week SMS text message group (aOR 1.25, 95% CI 1.00-1.56) were more likely to complete the screening test. Within the Cologuard-only model, patients in the 3-week SMS text message group (aOR 7.01, 95% CI 1.96-25.07) and the 6-week SMS text message group (aOR 5.75, 95% CI 1.53-21.61) had significantly higher odds of CRC screening completion. CONCLUSIONS: The findings highlight that moderate-frequency SMS text messaging reminders can effectively increase CRC screening rates in FQHCs; however, critical factors include the timing and frequency of these reminders. The 3-week intervention was associated with improved screening uptake, whereas the 6-week theory-informed intervention did not demonstrate a significant advantage over the control group, potentially reflecting a ceiling effect or message fatigue associated with more frequent messaging. Additionally, the study highlights unique screening patterns that contradict previous literature, underscoring the importance of a tailored approach for vulnerable communities.
15 May 2026
Read appraisal →JMIR formative research
Automatic Speech Recognition and Large Language Models for Multilingual Pathology Report Generation: Proof-of-Concept Study
BACKGROUND: Accurate transcription of pathology gross examination dictation is important for clinical documentation, but multilingual dictation remains challenging in settings where clinicians mix Chinese and English while final pathology reports are written in English. OBJECTIVE: This study aimed to evaluate whether a Whisper-based automatic speech recognition (ASR) pipeline guided by contextual system messages and combined with open-source large language models (LLMs; Qwen2:72b, Llama3.1:70b, Gemma2:27b) could improve multilingual (Chinese-English) pathology dictation transcription accuracy and generate clinically appropriate English gross description reports. METHODS: We conducted a controlled proof-of-concept study using 125 simulated mixed Chinese-English pathology gross examination audio recordings created by physicians or pathologists. Audio recordings were transcribed using Whisper ASR with and without a contextual system message. The ASR transcripts were then converted into English gross description reports using 3 open-source LLMs: Qwen2:72b, Llama3.1:70b, and Gemma2:27b. Outcomes included character error rate, Bilingual Evaluation Understudy, Recall-Oriented Understudy for Gisting Evaluation (ROUGE)-1, ROUGE-2, ROUGE-L, Metric for Evaluation of Translation with Explicit Ordering, pathologist Win-Tie-Lose rankings, report-level error categories, inference time, and interrater agreement. RESULTS: The ASR contextual system message reduced the mean character error rate from 0.344 (SD 0.176; 95% CI 0.313-0.375) to 0.066 (SD 0.100; 95% CI 0.048-0.084; P<.001). Qwen2:72b achieved the highest automated metric scores, including a Bilingual Evaluation Understudy of 0.644 (SD 0.307), ROUGE-1 of 0.866 (SD 0.163), ROUGE-2 of 0.771 (SD 0.235), ROUGE-L of 0.842 (SD 0.178), and Metric for Evaluation of Translation with Explicit Ordering of 0.805 (SD 0.214). Pathologist-coded total error rates were 16.8% (21/125) for Qwen2:72b, 45.6% (57/125) for Llama3.1:70b, and 92.8% (116/125) for Gemma2:27b. The exact agreement between the 2 pathologists across full ranking categories was 76.8% (96/125; Cohen κ=0.668), and agreement on the top-ranked model or tied top group was 81.6% (102/125; Cohen κ=0.722). CONCLUSIONS: In this proof-of-concept evaluation, contextual prompting improved ASR transcription accuracy, and Qwen2:72b generated the most accurate English pathology reports among the evaluated LLMs. However, the study used simulated audio recordings, a local vocabulary prompt, and report-level rather than term-level clinical annotation. LLM-generated reports should therefore be considered draft documentation requiring pathologist verification, and prospective validation in real clinical workflows is needed before clinical deployment.
15 May 2026
Read appraisal →Blood advances
Novel use of a hash-based tokenization for sharing sickle cell disease data without sharing protected health information
Efforts to improve care for people living with sickle cell disease (SCD) have led to the development of several registries; however, many are dependent on time-limited funding and lack coordination. Consequently, existing data sets are fragmented and do not provide the comprehensive, longitudinal insights achievable through well-integrated registries. Although relevant data exist within electronic medical records across institutions, aggregation is limited by poor interoperability, inconsistent use of common data elements, and poor translation of natural language into codified data. These barriers hinder population-level research and contribute to gaps in understanding the lifelong progression of SCD. Creating a new common data system risk losing years of valuable data, highlighting the need to optimize existing data resources. This study aimed to develop a privacy-preserving method to securely link 3 of the largest SCD data collection efforts in the United States. Conducted at the University of Alabama at Birmingham Lifespan Sickle Cell Center, the study leveraged institutional review board-approved access to the Sickle Cell Data Collection project, the American Society of Hematology Research Collaborative Data Hub, and the Globin Research Network for Data and Discovery. Identity tokens were generated and hashed using Secure Hash Algorithm (SHA)-256 to enable secure linkage without sharing protected health information. A total of 8026 records were identified across the 3 registries. Deterministic matching of hashed tokens identified 1080 unique individuals appearing in at least 2 data sets. This study demonstrates the first privacy-preserving linkage of multiple SCD registries. Secure data integration enhances interoperability and enables richer longitudinal analyses critical for advancing SCD research and treatment development.
15 May 2026
Read appraisal →Journal of medical systems
When Chatbots Become Agents: The Next Phase of Healthcare AI
Healthcare artificial intelligence (AI) has moved beyond answering medical questions. In early 2026, OpenAI, Anthropic, and Google launched agentic systems that retrieve evidence, use tools, and execute multi-step workflows. These systems can integrate information across multiple clinical knowledge domains within a single workflow, but the evidence base for evaluating their reliability in clinical practice has not kept pace with deployment. This perspective argues that healthcare is adopting a technology evolving faster than anyone can evaluate, and that a brief window exists to shape integration before the architecture becomes entrenched. These systems share common architectural principles, yet only 19 prospective trials exist among over 4,600 clinical large language model studies, none evaluating agentic pipelines. The architecture is locking in, professional roles are being redefined, and builders hold a structural information advantage. Transparency, adaptive governance, and broader participation are needed before these choices become irreversible, and to ensure that the boundary between human clinical judgment and automated capability is drawn in a way that strengthens both. Every stakeholder in healthcare must engage with this shift now.
15 May 2026
Read appraisal →Journal of cross-cultural gerontology
Mapping the Use of Information and Communication Technologies in Health Interventions for Older Adults: A Rapid Review
This rapid review mapped how information and communication technologies (ICTs), particularly digital health applications, are being used within healthcare services for adults aged 55 years and older. Following the Joanna Briggs Institute methodology and guided by the Population-Concept-Context framework, studies published between 2016 and 2024 were identified and reported in accordance with the PRISMA extension for scoping reviews, based on a previously published protocol registered on the Open Science Framework. Nineteen studies were included after critical appraisal. The findings indicate that ICT use is most frequently associated with improved accessibility and coordination of healthcare services, support for health professionals' training and care management practices, and positive health-related outcomes for older adults, particularly in the domains of mental health, social participation, and self-care. However, the evidence remains heterogeneous and often limited in its reporting of design and implementation processes, highlighting the need for further context-sensitive research that addresses ethical, equity, and implementation considerations in the use of digital health technologies for older populations.
14 May 2026
Read appraisal →PloS one
TDGN: A text-guided dual-gated network for multimodal sentiment analysis
Multimodal Sentiment Analysis (MSA) aims to interpret emotions by integrating textual, acoustic, and visual information. However, the heterogeneous quality and weak correlation among nonverbal modalities often lead to unstable alignment and ineffective fusion. To address these challenges, we propose a Text-Guided Dual-Gated Network (TDGN) that introduces a hierarchical gating and text-anchored contrastive learning framework. During the alignment phase, a Text-Anchored Gated Attention (TGA) module employs text as a semantic anchor to guide fine-grained alignment between audio and visual modalities while suppressing noise and emphasizing salient emotional cues. For the fusion stage, a Dual-layer Gated Fusion (DGF) module performs intra-modal gating to refine modality-specific features and inter-modal gating to dynamically balance cross-modal contributions. Furthermore, we introduce a Text-Anchored Contrastive Learning (TACL), which constructs contrastive targets based on textual similarity anchors, ensuring the fused features maintain both modal consistency and modal diversity. Extensive experiments on the CMU-MOSI and CMU-MOSEI benchmarks demonstrate that TDGN achieves state-of-the-art performance, achieving higher accuracy and robustness. Ablation studies and visualization further validate the effectiveness of each component.
14 May 2026
Read appraisal →Journal of medical systems
VR-Based Physical Activity and Self-Management Interventions for Glycemic Control and Behavioral Outcomes in Type 2 Diabetes: A Systematic Review and Meta-Analysis
A growing body of original research has recently investigated virtual reality (VR)-based interventions for physical activity and self-management, yet yielded inconsistent results. This systematic review and meta-analysis aims to evaluate the effectiveness of VR-based physical activity and self-management interventions in improving glycemic control and modifying health-related behaviors among patients with type 2 diabetes mellitus (T2DM). A total of 10 studies involving 1,005 adults with T2DM were included, comprising 5 randomized controlled trials (RCTs), 1 non-randomized controlled trial (nRCT), and 4 self-controlled before-after studies. Of these, participants in the intervention groups received VR-based interventions, including immersive or interactive VR programs for diabetes self-management education or physical activity promotion, with most intervention durations ranging from 8 to 24 weeks. Control groups received traditional approaches, including routine care, in-person diabetes self-management programs, and content-equivalent 2D websites. The pre-post intervention analysis revealed that glycated hemoglobin (HbA1c) levels decreased significantly post-intervention (mean difference (MD) = -0.50, 95% CI: -0.92 to -0.08, p = 0.02). While VR-based physical activity and self-management interventions did not demonstrate superior efficacy to traditional intervention methods in glycemic control or behavioral outcomes, their unique advantages - including enhanced adherence, greater accessibility, and reduced spatiotemporal constraints - position them as a viable alternative for diabetes management. Nevertheless, the field requires more rigorously designed RCTs to establish efficacy and a systematic exploration of VR implementation protocols to maximize therapeutic benefits.
14 May 2026
Read appraisal →JMIR nursing
From Concept to Practice: Lessons From the Balanced Nursing Teams Decision-Support System
The global nursing workforce crisis demands a shift from reactive staffing to strategic workforce optimization through data-driven decision-support systems. This viewpoint paper reflects on the development and attempted implementation of the balanced nursing teams system, a decision-support tool integrating approximately 250 data points-of which roughly 150 are extracted from existing organizational systems (human resources, scheduling, electronic health records, quality registries) through flexible import mechanisms, and the remainder collected through a built-in 360-degree staff survey with automated analysis-across 10 domains to evaluate nursing team balance between capacity, performance, and outcomes. Following crowdfunding by 18 Belgian health care organizations, balanced nursing teams were implemented across 8 diverse settings (home health care, general hospitals, academic centers) between 2019 and 2023. Using the Human-Organization-Technology fit framework, we analyze why evidence-informed, organization-endorsed digital innovations struggle to achieve adoption. Our analysis reveals 3 interdependent barrier categories: technological fragmentation (vendor lock-in, legacy systems, prohibitive integration costs), organizational siloing (Chief Nursing Officers [CNOs] lacking budgetary authority, nursing framed as peripheral to strategic priorities), and managerial hesitance (fear of punitive data use, cognitive overload from staffing crises). These barriers were worsened by the substantial data-integration burden that the system's breadth imposed on organizations with limited digital maturity. Critically, only one site (ie, a nurse-led home health care organization where leadership held both strategic authority and resource control) achieved sustained implementation. This contrast demonstrates that workforce optimization through data depends not on software maturity alone, but on achieving simultaneous fit across human, organizational, and technological domains. We argue that the persistent marginalization of nursing leadership within hospital governance structures represents the fundamental barrier to digital transformation in nursing workforce management. The urgency paradox is striking: while nursing represents health care organizations' highest operational cost and most direct patient interface, workforce optimization tools are consistently deprioritized in favor of regulatory compliance systems and billing infrastructure. Bridging this gap requires systemic investment in nursing leadership authority, data interoperability standards, and recognition that data-driven workforce decisions are strategic imperatives rather than operational luxuries.
14 May 2026
Read appraisal →JMIR aging
Self-Identified Age Cohorts and Personal Experience Sharing in Pseudonymous Online Spaces: Natural Language Processing Analysis of Reddit
BACKGROUND: While older adults' social media use has been widely studied for its instrumental benefits, such as accessing health information or maintaining family ties, research has largely focused on identity-based platforms that mirror offline social networks, leaving pseudonymous, interest-driven environments such as Reddit underexplored. Although older adults actively participate in these spaces to share personal narratives and engage beyond their existing social roles, the literature has yet to center their own voices, with most existing work focusing on caregivers or younger users discussing older adults rather than older adults speaking for themselves. OBJECTIVE: Integrating a life course perspective with the concept of subjective age cohort, this study examines how Reddit users who explicitly self-identify as being in their 50s, 60s, or 70s engage in self-expression and personal experience sharing within pseudonymous online spaces. METHODS: Using data collected via the Reddit application programming interface, this study analyzed posts and comments from 848 self-identified older Reddit users, including 488 future older adults in their 50s (57.5%) and 360 current older adults in their 60s and 70s (42.5%) identified through age-based flairs in the subreddit r/AskOldPeople. A fine-tuned large language model (BLOOM-560m; F1-score=0.96) classified 4,055,275 sentences into three personal experience domains: (1) health and wellness, (2) personal relationships and identity, and (3) professional and financial life. Chi-square analysis compared domain distributions across age groups. BERTopic topic modeling identified thematic patterns within each domain. RESULTS: Of 569,107 personal experience sentences identified, personal relationships and identity comprised the largest share (n=268,212, 47.1%), followed by professional and financial (n=186,768, 32.8%) and health and wellness (n=114,127, 20.1%). Chi-square analysis revealed significant between-group differences (χ22=34.7; P<.001): current older adults shared proportionally more about health and wellness, whereas future older adults shared more about relationships and identity. Topic modeling further revealed qualitatively distinct emphases within each domain. Future older adults' posts frequently discussed menopause, depression, and friendships, while current older adults more frequently addressed chronic illness, aging, and financial security. Both groups used pseudonymity to disclose sensitive or stigmatized topics. CONCLUSIONS: This study demonstrates that older adults' personal experience sharing on pseudonymous platforms is shaped by the simultaneous interplay of life course stage, age cohort background, and platform affordances, rather than by any single factor. Pseudonymous environments such as Reddit enable disclosures of health vulnerabilities, contested social identities, and emotional experiences that are structurally discouraged on identity-based networks, and this has direct implications for how digital inclusion is conceptualized and practiced. Beyond operational literacy, supporting older adults' expressive inclusion in interest-based, pseudonymous online communities represents a meaningful yet underaddressed dimension of digital participation policy.
14 May 2026
Read appraisal →PloS one
Framing artificial intelligence in Chilean digital press before and after the launch of ChatGPT: From concern to optimism
This article examines the evolution of AI framing in Chilean national digital media before and after the launch of ChatGPT in 2022. Through topic modeling and emotion analysis of 1,466 articles published by six media outlets over periods that begin between 2000 and 2014, depending on the outlet, and extend in all cases until 2024, this study identifies significant thematic and affective changes. In the pre-ChatGPT period, coverage combined recognition of AI's capabilities with concerns regarding labor displacement, governance, and human identity, frequently referencing global corporations and state institutions. In the post-ChatGPT period, topics became narrower in scope, with fewer actors and greater emphasis on universities and cultural organizations implementing AI. The focus shifted from the development of AI systems to their applications, predominantly framed in positive terms. At the same time, negative sentiment was largely confined to epistemological uncertainty. This transformation aligns with national surveys indicating growing public optimism, diverges from Global North debates centered on regulation and privacy, and converges with East Asian framings that are oriented toward innovation and creativity. These findings suggest that the launch of ChatGPT coincided with a reinforcement of a socio-technical imaginary of AI as inevitable and beneficial, yet potentially limiting the diversity of public debate.
14 May 2026
Read appraisal →JMIR medical informatics
Expert Evaluation and Consensus on GPT-4o Summaries of Clinical Letters: Validation and Results of the Framework and Implementation of AI Tools Project
BACKGROUND: Large language models (LLMs) are increasingly used to summarize clinical documents; yet, automated metrics often inadequately capture clinical relevance and safety. In the initial phase of the "Framework and Implementation of AI Tools," an expert-driven, cocreated evaluation methodology was established to assess LLM-generated discharge letter summaries, combining prompt design considerations with intuitive expert appraisal. OBJECTIVE: This study aimed to quantify expert agreement and interrater reliability on LLM summaries of discharge letters, identify frequent and clinically relevant errors, and evaluate practical implications for integrating LLMs into documentation workflows. METHODS: Thirty expert-curated synthetic Dutch discharge letters were summarized. Thirty-one clinicians from Flemish care settings (1 university hospital, 2 private hospitals, and 2 general practice circles) evaluated the summaries. The evaluation framework consisted of 61 binary layout items assessing whether required sections and formatting were correctly present, 33 content items (correct or complete vs incorrect, subcategorizing missing, irrelevant, and hallucinated information), a 4-point global quality rating, and an open comment. Statistical analyses included descriptive statistics, mixed effects ordinal regression on the global score, consensus (agreement per question or letter) percentages, interrater reliability (Cohen κ, intraclass correlation coefficient [ICC], Fleiss κ, and prevalence index), and thematic synthesis of comments. RESULTS: Layout adherence was high (88%), especially in the conclusion section. The positive response rate for content was overall moderate (78%), with the best performance observed in the medical history section and the lowest performance observed in the medication section, which also showed the highest rate of hallucinations and the weakest interrater consensus. Across all sections, missing information was the most common error. Nearly 70% of global ratings were "good" or "very good." Higher positive response rates for content predicted better global scores (β=.079; P<.001), while layout and participant specialty were not relevant to global scoring. Consensus was high for the layout questions (median 96.8%, IQR 90.2%-100%) and somewhat lower for content (median 83.9%, IQR 67.7-96.8), with the lowest agreement in the medication section. Interrater agreement was moderate (median Cohen κ=0.36, IQR 0.29-0.43; range 0.07-0.56), but overall reliability was high (ICC 0.945, 95% CI 0.942-0.948), indicating strong consistency at the global level despite interrater variability. The prevalence index demonstrated that high ICC values were partly driven by the strong prevalence of affirmative responses in layout items, while content items showed more balanced distributions and lower agreement. CONCLUSIONS: Our framework offers a robust approach for evaluating LLM-generated discharge summaries, balancing usability and clinical relevance. Semantic integrity, especially regarding medication details, was identified as a key vulnerability. Perceived overall quality was driven by a positive response rate for content. High ICC values for global score, with lower item-level agreement lead toward the need for clearer, context-specific prompts and standardized evaluation criteria to reduce interrater variability. Human oversight and targeted automated checks for omissions and hallucinations are essential for safe clinical deployment.
14 May 2026
Read appraisal →PloS one
Extracting user profile via large language models and ontologies
Extracting user-specific profiles that include general personal information, such as hobbies, occupation, or age, is a valuable asset for systems like recommendation engines or personalized chatbots. Currently, most approaches moved toward exploiting the capabilities of large language models (LLMs) on rich languages while less-resourced languages still present significant opportunities for exploration and improvement. In this research, we present a multi-step approach for profile extraction from dialogue systems in Persian as a use case. Through an extensive set of experiments and analyses on various models, we show that LLMs struggle with these languages due to limited language-specific resources and complex linguistic structures. To address this, we propose a hybrid method that combines techniques such as slot filling, in-context learning, and ontology-based inference. Our final results demonstrate a significant improvement over current state-of-the-art models, including LLMs with few-shot examples and even their fine-tuned version. Our method achieves an F-score of 90.46, outperforming GPT-4o and Llama-3-70B by an absolute difference of 17.08 and 24.51 respectively. Our system can also detect inconsistencies in presented information in which our performance substantially exceeds the best performance of GPT-4o and Llama-3-70B with an accuracy of 92%. It is 21% absolutely better than GPT-4o and 47% better than Llama.
13 May 2026
Read appraisal →JMIR medical informatics
Natural Language Processing for Automated Classification of Cleft and Craniofacial Procedures From Operative Notes: Model Development and Feasibility Study
BACKGROUND: The accurate classification of operative notes is essential for surgical outcomes research; however, CPT code classification is notoriously nonspecific for many procedures. In such situations, the operative note (or "dictation") must be reviewed manually, a process that is labor-intensive and unsustainable. Natural language processing demonstrates tremendous potential for improving the efficiency and accuracy of procedure classification from unstructured operative notes. To date, it remains unexplored whether natural language processing can reliably differentiate between complex, multicomponent procedures, such as those involved in the care of cleft lip or palate and craniofacial anomalies. OBJECTIVE: This study aims to develop and evaluate a machine learning framework for the automated classification of operative notes for cleft and craniofacial procedures. METHODS: This single-institution, retrospective observational study used operative notes from patients undergoing cleft and craniofacial procedures at a single academic medical center from 2016 to 2024. Each note in the database had been manually classified previously. Notes were preprocessed and vectorized using term frequency-inverse document frequency. A One-vs-Rest classification framework with random forest as the base classifier was developed to categorize procedures at 3 levels: primary procedure type (cleft lip repair, alveolar bone grafting, cleft palate repair, velopharyngeal insufficiency correction, rhinoplasty, and other), procedural subtype (primary vs revision), and specific surgical technique used (eg, Fisher, Mulliken, or rotation-advancement technique for cleft lip repair). Each hierarchical level was developed and evaluated using cross-validation. To improve procedural subtype classification for classes with few samples, synthetic notes were added to the dataset. Area under the receiver operating characteristic curve (AUC), an area under the precision-recall curve, micro- and macro-averaged F1-scores, and Hamming loss were used to assess model performance. RESULTS: The dataset comprised 630 operative notes from 311 pediatric patients undergoing cleft and craniofacial procedures between 2016 and 2024, with a mean age of 3.75 (range 0-19) years. The primary classification model achieved strong performance in distinguishing procedure types with an AUC of 0.93 (SD 0.04), area under the precision-recall curve of 0.84 (SD 0.05), micro-averaged F1-score of 0.88 (SD 0.02), a macro-averaged F1-score of 0.84 (SD 0.03), and a Hamming loss of 0.04 (SD 0.01). Secondary classifiers achieved AUC scores of 1.0 (SD 0.00) for cleft lip revision classification but failed to discriminate between alveolar bone grafting primary and revision procedures (AUC 0.49, SD 0.02). Tertiary classifiers for surgical technique identification showed AUC scores of 0.88 (SD 0.03), 0.89 (SD 0.03), and 0.89 (SD 0.09) for cleft lip, cleft palate, and velopharyngeal insufficiency repair techniques, respectively. CONCLUSIONS: This pilot study demonstrates that machine learning approaches can automate the classification of pediatric craniofacial operative notes across multiple levels of procedural detail. The implementation of such systems could significantly reduce the administrative burden related to surgical research, operations, and quality improvement.
13 May 2026
Read appraisal →JMIR human factors
The Quality and Characteristics of Digital Mental Health Apps: Mixed Methods Study
BACKGROUND: There are around 20,000 mental health apps available in app stores. The Organisation for the Review of Care and Health Apps (ORCHA), a United Kingdom digital health compliance company, has assessed a number of digital mental health apps with regard to their quality, professional and clinical assurance, data privacy, and user experience. This study analyzes the data that were collected by ORCHA when they assessed mental health apps. OBJECTIVE: This study aimed to examine the characteristics of mental health apps regarding their quality, target users, features, underpinning evidence, and data privacy. METHODS: A dataset comprising ORCHA Baseline Review assessments of over 2000 digital health apps, including 436 mental health apps, was used. This study uses exploratory data analysis to gain insight into the quality and characteristics of mental health apps. Methods such as descriptive and inferential statistics, k-modes clustering, and association rule mining were used to explore the quality of mental health apps as well as reveal insights into the different cost types, target users, app features, data types, and evidence of app content. RESULTS: Information provision, data capture, and data sharing were the most common features within the 436 mental health apps. The examined apps primarily targeted the following groups: adults (n=229, 52.5%), everyone (n=184, 42.2%), and teens (n=135, 31%). The cost of apps has not been linked to the quality of mental health apps, although paid apps or apps with in-app purchases may include additional services. Indicated user acceptance or benefit is the most common type of evidence provided by these mental health apps. A total of 241 (55.3%) apps included a qualified professional in app development, and 251 (57.6%) apps provided evidence within the app that the developer validated any guidance with relevant reliable information sources or references. Usage data and email were the most commonly collected data types. Association rule mining showed that email, IP address, name, and usage data are often co-collected by the same apps. K-modes cluster analysis showed that mental health apps can be categorized into 2 clusters, where one cluster of apps (n=182, 41.7%) collected more data than apps in the other cluster. CONCLUSIONS: Mental health apps are commonly targeted for everyone to use, but many apps are targeted toward teens or adults. Our study suggests that many publicly available mental health apps did not take the precautions (such as the involvement of appropriate health professionals, literature references, or conducting tests) to ensure that their content is valid and research based. Greater effort on behalf of mental health app developers is needed to ensure that the public is provided with high-quality apps. Moreover, our study indicates that the mental health apps that collect more data tend to score better on the ORCHA Baseline Review assessment.
13 May 2026
Read appraisal →JMIR research protocols
Evaluating Patient-Entered Electronic Health Data as a Strategy to Improve Quality of Care in a Diabetes Clinic: Protocol for a Randomized Controlled Trial
BACKGROUND: Quality care for pediatric type 1 diabetes (T1D) requires frequent, multidisciplinary visits. Technological and clinical innovation have led to changes in T1D management, resulting in increasing data exchange required during these visits. Capturing comprehensive personal health and diabetes-related information discretely and integrating it into the clinical workflow is critical for optimal T1D care but is time-consuming. Time spent on data transfer often results in less time for holistic care and can lead to unmet needs for patients, families, and health care providers, as well as increased time pressures in clinic. To address this, the Children's Hospital of Eastern Ontario developed a caregiver proxy-reported questionnaire distributed via the MyChart patient portal, allowing families to input care information ahead of visits with the aim of dedicating more clinic time to personalized care. The launch of this tool, which integrates caregiver-entered information into the physician's documentation workflow, brings the opportunity to systematically evaluate its impact on care quality and efficiency, with potential implications for broader adoption. OBJECTIVE: Our objective is to evaluate the impact of a caregiver proxy-reported, electronic health record-integrated preclinic questionnaire (MyChart questionnaire) on the quality of care in a pediatric diabetes clinic, through measurement of its impact on caregiver-perceived quality of care compared to the standard of care using 2 validated measures of care quality. We also aim to explore the impact of the intervention on glycemic control and visit efficiency. METHODS: We conducted a single-center, parallel-group randomized controlled trial designed for 222 children with T1D. Participants were randomly allocated in a 1:1 ratio to either the intervention (MyChart questionnaire) or standard care. Our primary outcome is caregiver-perceived quality of care, as measured by the Patient's Evaluation of the Quality of Diabetes Care at 8 months, administered with caregivers serving as proxy respondents for patients. Secondary outcomes are the Patient's Evaluation of the Quality of Diabetes Care at 4 months and Perceived Quality of Medical Care at 4 and 8 months. Tertiary outcomes include glycemic control and physician-reported visit efficiency at 4 and 8 months. Analysis of covariance models will be used to assess changes between baseline and postintervention outcomes across treatment groups. RESULTS: Recruitment for this study began in April 2023 and was completed in February 2024, with a total of 139 participants enrolled. Data collection has concluded, and the first results are expected in the spring of 2026. CONCLUSIONS: This study is the first randomized trial to assess the impact of a caregiver proxy-reported, electronic health record-integrated, preclinic questionnaire distributed via a patient portal on caregiver-perceived quality of care in a pediatric care setting. The results will guide changes in health service infrastructure and delivery to enhance comprehensive data capture and improve care quality within and beyond pediatric T1D.
13 May 2026
Read appraisal →PloS one
A modular deep learning architecture for interpretable disease prediction across tabular clinical and biometric datasets
Accurate disease prediction using clinical datasets is essential for improving early diagnosis and clinical decision-support systems; however, many existing deep learning approaches are disease-specific, computationally intensive, and difficult to generalize across heterogeneous biomedical datasets. This study addresses this challenge by proposing a unified and dataset-aware deep learning framework that enables accurate and interpretable disease prediction across diverse clinical datasets. The framework adopts a modular architecture that selects appropriate models based on dataset characteristics such as feature dimensionality, sample size, and class imbalance. It integrates multiple deep learning architectures, including MLP, one-dimensional CNN, FT-Transformer, autoencoder-based classifiers, and ensemble strategies. Robust preprocessing, fold-safe feature selection, and nested cross-validation are incorporated to ensure reliable performance evaluation. The framework is evaluated on three heterogeneous benchmark datasets: the UCI Heart Disease dataset (303 samples, 13 clinical features), the PIMA Indians Diabetes dataset (768 samples, 8 metabolic features), and the Parkinson's disease voice dataset (195 recordings, 22 acoustic features). Experimental results demonstrate competitive predictive performance relative to classical baselines across the diverse tasks. The FT-Transformer + autoencoder ensemble achieved an AUC of 0.8980 (±0.0483) for heart disease prediction, while the CNN + Autoencoder ensemble obtained an AUC of 0.8451 (±0.0270) for diabetes classification. For Parkinson's disease detection, the MLP achieved an AUC of 0.7538 with perfect specificity. Overall, all models achieved AUC values comparable to ML baselines. The study contributes a scalable and interpretable deep learning framework that improves reliability, generalization, and practical applicability for multi-disease prediction in real-world healthcare environments.
13 May 2026
Read appraisal →Journal of medical Internet research
Governing Patient-Facing AI-Generated Video in Digital Health: A Risk-and-Ethics Matrix for Deployment, Monitoring, and Change Control
In this Viewpoint, we argue that patient-facing high-fidelity artificial intelligence (AI)-generated video requires governance that is operational, life cycle based, and embedded in existing institutional review pathways rather than limited to predeployment checks alone. Patient-facing high-fidelity AI-generated video-synthetic or substantially AI-mediated video that presents realistic human likeness, voices, or clinical communication cues-is rapidly entering patient education and clinical communication. We propose a risk-and-ethics matrix that combines residual clinical risk (likelihood × severity after mitigations) with an ethical alignment score that operationalizes autonomy, beneficence, nonmaleficence, and justice to yield actionable dispositions (encourage, permit with oversight, restrict or redesign, or prohibit). The framework links each disposition to dossier-based review, minimum controls, and postdeployment monitoring triggers-focused on measurable outcomes (eg, comprehension, content-attributable follow-up burden, incidents and complaints, and equity gaps) as well as provenance and change control-to support auditable, revisitable decisions over the system life cycle.
13 May 2026
Read appraisal →Scientific reports
Uncovering advanced transfer learning strategies for deep neural networks in natural language processing
In the rapidly evolving landscape of Natural Language Processing (NLP), transfer learning has emerged as a game-changing methodology, fundamentally altering how machine learning models are trained and deployed. The study at hand dives deep into the intricacies of transfer learning techniques, specifically focusing on their application in complex deep learning architectures within the NLP domain. By exploring a variety of architectural designs, fine-tuning methodologies, and alternative training paradigms, we aim to demystify the optimal strategies for harnessing the power of pre-trained models. To quantify the effectiveness of these approaches, we conducted a comprehensive series of experiments targeting key NLP tasks, such as text classification and language generation. Transfer learning significantly accelerated training and improved model accuracy, highlighting its practical advantages in NLP tasks. This dual benefit underscores the immense potential of advanced transfer learning techniques, making them an indispensable tool for future NLP applications. By implementing state-of-the-art transfer learning methodologies, companies can offer faster and more accurate NLP solutions, aligning perfectly with the ethos of "Business Digitalized" and catering to a broad spectrum of client needs in the market.
13 May 2026
Read appraisal →PloS one
Systematic evaluation of the DeepSeek large language model for clinical diagnostic reasoning
BACKGROUND: Artificial intelligence (AI) is undergoing an era of transformative advancement, particularly through the emergence of Transformer-based large language models (LLMs). While these systems demonstrate strong reasoning and generalization capabilities, their clinical applicability, particularly in emergency and critical care decision-making, remains underexplored.. In time-sensitive settings, diagnostic reasoning must align rigorously with evidence-based standards and ensure the relevance of timing to clinical decisions. OBJECTIVE: This study aims to provide a preliminary evaluation of the decision-support performance of the DeepSeek model in acute medical scenarios. We systematically evaluate its diagnostic reasoning, temporal consistency of recommendations, and adherence to evidence-based critical care protocols using standardized case-based assessments. METHODS: Twenty-nine representative clinical cases were extracted from the Merck Manual of Diagnosis and Therapy, a widely used medical reference providing standardized case descriptions. The model's outputs were evaluated across four decision-making dimensions: differential diagnosis, diagnostic testing, final diagnosis, and management planning. Human raters scored each response for accuracy, and multivariable linear regression was applied to assess associations between performance and case parameters (age, gender, and Rapid Emergency Medicine Score [REMS]). RESULTS: DeepSeek achieved an overall mean accuracy of 82.9% (95% CI: 80.2-85.6%) across all cases. Accuracy peaked in final diagnosis (97.7%), but declined in differential diagnosis (73.0%). Model performance showed no significant variation across demographic or severity strata. CONCLUSIONS: DeepSeek shows promising performance in structured case-based diagnostic tasks, particularly in confirmatory diagnostic reasoning. However, its early-stage reasoning and handling of ambiguous cases require enhancement. Future studies using larger and more diverse clinical datasets are needed to further evaluate the model's robustness and potential clinical applicability.
13 May 2026
Read appraisal →Scientific reports
Automated phenotyping of ophthalmologic diseases from routine medical records using small language models and the human phenotype ontology (HPO)
Automated phenotyping in ophthalmology requires accurate standardization of clinical terms to facilitate interoperability and research. This study evaluates the suitability of the human phenotype ontology (HPO) for automated extraction of ophthalmic phenotypes from narrative documentation. We developed a locally operated AI pipeline combining text segmentation and negation detection based on a small language model (PHI-4) with a dense retrieval approach using an augmented multilingual HPO catalog. Synonyms were incorporated into the HPO during training on anonymized consecutive physician letters. To validate the pipeline, 175 anterior segment and fundus descriptions from randomly picked medical records were manually annotated with HPO terms as ground truth. Overall, 342 HPO terms were identified manually (on average 2.53 terms per document), with 341 retrieved by the pipeline (on average 2.52 terms per document). Performance metrics showed a median Jaccard similarity of 0.67, precision of 0.83, recall of 0.82, and F1 score of 0.80. These results demonstrate that our AI pipeline effectively extracts standardized HPO terms from free-text ophthalmic findings. Integration of this pipeline into clinical information systems may enhance data interoperability and reduce manual coding workload in ophthalmology practices in the future.
12 May 2026
Read appraisal →Journal of medical Internet research
Barriers and Facilitators to Patient Acceptance of Artificial Intelligence in Health Care: Systematic Review
BACKGROUND: Artificial intelligence (AI) in the domain of health care is increasing in prominence. Acceptance is an indispensable prerequisite for the widespread implementation of AI. OBJECTIVE: This study aimed to explore barriers and facilitators influencing patients' acceptance of AI. METHODS: We conducted a systematic review following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Nine databases, including PubMed, Web of Science, and Embase, were comprehensively searched from inception to December 23, 2025. We included qualitative, quantitative, and mixed methods studies investigating adult patients' attitudes toward medical AI. Two researchers independently screened records, extracted data, and appraised methodological quality using the Mixed Methods Appraisal Tool. Following the Joanna Briggs Institute convergent integrated approach, data synthesis was guided by integrating the Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) and the Theoretical Domains Framework (TDF). Factors were mapped to behavior change techniques (BCTs) and evaluated for practical feasibility using the Affordability, Practicability, Effectiveness and cost-effectiveness, Acceptability, Side-effects/safety, and Equity criteria. RESULTS: A total of 61 studies met the inclusion criteria out of 7452 search results. Study designs included qualitative (n=20), quantitative (n=35), and mixed methods (n=6). Performance and effort expectancies were the primary determinants of acceptance. Major barriers comprised perceived operational complexity, lack of algorithmic trust, reduced interpersonal interaction, privacy vulnerabilities, and high costs. Facilitators included transparent data governance, interpretability of AI decisions, improved clinician-patient communication, and human-centered design. Education level and disease severity emerged as key moderating variables. Through UTAUT2-TDF mapping, we identified 25 distinct BCTs (6 high, 14 medium, and 5 low feasibility) and formulated 40 actionable intervention strategies. CONCLUSIONS: This study innovatively integrates the UTAUT2 and TDF frameworks to evaluate patient acceptance of medical AI. Unlike existing reviews that predominantly evaluate isolated psychosocial factors or purely technical attributes, this transtheoretical approach differentiates itself by merging technology adoption mechanisms directly with behavioral drivers. Consequently, it contributes to the field by systematically identifying multilevel factors influencing acceptance, including performance expectancy, effort expectancy, and ethical security, and translating these into 40 actionable BCTs. In real-world clinical practice, these findings provide a feasible, prioritized blueprint for clinicians and administrators to design patient-centered interventions, enhancing the clinical integration and long-term effectiveness of medical AI.
12 May 2026
Read appraisal →Journal of medical Internet research
Why We Need Patients and Community at the Center of AI Health Communication Research
Holderried and colleagues tested whether artificial intelligence (AI)-generated, patient-centered information can help people understand what they need to do after being discharged from a hospital. Participants demonstrated stronger comprehension when they viewed the simplified, patient-centered information rather than a standard letter. This work adds to the available early-phase evidence of AI supporting hospital discharge communication. To meaningfully progress this area of research, we now need to carefully consider how to enhance the design and evaluation of patient-facing AI health communication tools. In this commentary, we argue that equity and consumer engagement remain underrepresented in studies on patient-facing AI health communication tools and describe possible approaches to address this issue.
12 May 2026
Read appraisal →PloS one
Gated subspace alignment with drift compensation for parameter-efficient Class-Incremental Learning
Class-Incremental Learning (CIL) aims to enable models to continuously learn new categories while preserving existing knowledge and avoiding catastrophic forgetting. Although parameter-expansion architectures can alleviate task interference to some extent, the representations of previously learned classes often drift or degrade as the feature subspaces continuously evolve and expand, resulting in decreased recognition performance for old classes. To address this issue, we propose an efficient CIL method-Dynamic Gated Adapter for Subspace Alignment (DGASA). Based on a frozen pre-trained backbone, DGASA introduces lightweight adapters with attention-based gating for each task to construct task-specific subspaces, while dynamically fusing cross-task information via attention mechanisms. In addition, DGASA learns a linear mapping between the old and new subspaces to achieve consistent alignment of old class prototypes in the current subspace without accessing past data. Extensive experiments demonstrate that DGASA significantly improves classification accuracy and resistance to forgetting on multiple benchmark datasets, offering strong generalization and computational efficiency.
12 May 2026
Read appraisal →JMIR human factors
Persuasive Systems Design Features of Smartphone Apps for Psychosis: Systematic Review
BACKGROUND: It is unclear why some smartphone apps designed for people with psychosis are engaging, while others are not. One possible explanation is the apps' persuasive features and the operationalization and implementation of these features. OBJECTIVE: This systematic review set out to quantify and describe the persuasive features used in smartphone apps for psychosis, investigate whether there was any association between persuasive features and attrition or adherence rates. and document the quality of the included apps. METHODS: We searched electronic databases PsycINFO, PubMed, and Google Scholar for eligible papers published between the years of 2013 and 2025. Hand searches of reference lists were completed. Apps were selected if they were designed for people with psychosis and there were published empirical studies investigating the apps. Two reviewers from the review team (TT, JP, DK, and SD) independently screened papers and extracted data on adherence and attrition, as well as coded papers for evidence of persuasive features according to the persuasive systems design (PSD) model. Available data were synthesized descriptively and narratively. We attempted to access apps via app stores or by correspondence with the research team. RESULTS: We found 22 apps for psychosis, with 30 associated published papers. The persuasive features were as follows: personalization (18 apps); reminders (15 apps); suggestions (11 apps); tunneling and self-monitoring (10 apps); reduction (9 apps); liking (8 apps); social role (6 apps); rehearsal, praise, and similarity (5 apps); rewards (4 apps); simulation, real-world feel, and social learning (3 apps); surface credibility and normative influence (2 apps); and trustworthiness, social comparison, and social facilitation (1 app). Expertise, authority, third-party endorsements, verifiability, cooperation, competition, and recognition were present in zero apps. Features in the categories of primary task support and dialogue support were well represented, while social support and system credibility support were underused. It was found that there was no association between the number of persuasive features and attrition; an association between persuasive features and adherence could not be assessed. The quality of the apps could not be judged due to 20 of the 22 apps being inaccessible either through the research papers' authors or through app stores. CONCLUSIONS: Our findings indicate that in psychosis apps there is potential to include a broader range of persuasive features, which might maximize engagement. Psychosis apps may benefit from incorporating more features that leverage the persuasive impact of having users interact (social support) and incorporating features that emphasize system credibility and trustworthiness. Further studies could determine whether an increase in the number of persuasive features will impact app engagement and which features are most impactful in this context.
12 May 2026
Read appraisal →Physics in medicine and biology
Multi-model study of fast VMAT segment dose calculation with deep learning.
Objective.Deep learning (DL) methods enable photon dose calculation under two main coordinate representations: Beam's Eye View (BEV) and patient coordinates. We evaluate dose calculation accuracy and speed under these coordinate paradigms and with representative DL models within a unified dataset and pipeline, and introduce two lightweight models for fast photon dose calculation.Approach.Planning computed tomography (CT) scans and volumetric modulated arc therapy plans from 24 prostate cancer patients were used. Monte Carlo simulation generated 5940, 540, and 3053 segment doses for training (11 patients), validation (3), and testing (10), respectively. For BEV, we used a combination of convolutional neural network (CNN) and convolutional long short-term memory network (ConvLSTM) called CNN-ConvLSTM, a CNN-Mamba combination (CNN-Mamba), a transformer-based architecture (DoTA), and a cascaded 3D UNet (C3D). These were trained on CT and segment-projection BEV cuboids. For patient coordinates, the DeepDose individual segment dose prediction framework implemented with C3D (DeepDose-C3D) was trained on cropped CT volumes with four physical inputs. Segment and plan dose accuracy were assessed using local gamma passing ratesγPR(2%/3 mm and 1%/3 mm) and dose-volume histogram metrics. Dose calculation times (inference plus pre/post-processing) were measured on three different graphics processing unit (GPUs).Results.All five models achieved mean localγPRvalues⩾91.0% (2%/3 mm) for segment doses and⩾99.0% (1%/3 mm) for plan doses. Mean per-segment dose calculation times were 79, 67, 298, 490, 356 ms for CNN-ConvLSTM, CNN-Mamba, DoTA, C3D, and DeepDose-C3D, respectively. On the latest-generation GPU available, the corresponding per-plan (average 305 segments) dose calculation times were 5.5, 6.2, 33.6, 38.7, 35.4 s.Significance.Both BEV- and patient-coordinate DL methods achieved accurate photon plan dose calculation, with BEV-based approaches showing more robust segment performance. CNN-ConvLSTM and CNN-Mamba retain comparable accuracy at lower computational cost, enabling fast photon dose calculation.
12 May 2026
Read appraisal →Journal of medical Internet research
Automated Approaches of Text Simplification of Patient Education Materials: Scoping Review
BACKGROUND: Patient education materials (PEMs) often exceed the American Medical Association's (AMA) recommended sixth-grade reading grade level (RGL). While artificial intelligence (AI) offers potential for automated text simplification, concerns persist regarding linguistic quality, content fidelity, and the understandability of simplified PEMs by laypeople. OBJECTIVE: This scoping review maps existing evidence on automated language processing technologies for simplifying PEMs for laypeople. METHODS: Following the Joanna Briggs Institute (JBI) methodology and the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guideline, 5 bibliographic databases (Ovid MEDLINE, Embase, CINAHL, PsycInfo, and IEEE Xplore) were systematically searched from 2019 to May 2025, supplemented by reference screening and gray literature searches. Eligible sources were peer-reviewed empirical studies published in English that examined large language models (LLMs), AI-supported writing assistants, AI-based conversational agents, or AI-supported tools designed for automatic text simplification of PEMs. Targeted outcomes included linguistic quality (ie, linguistic comprehensibility, linguistic correctness) and content fidelity (ie, factual accuracy, factual completeness) of simplified PEMs. Excluded sources comprised rule-based systems, manual text simplification, non-laypeople target groups, and technology-focused performance metrics. Results were synthesized via thematic analysis across the domains of targeted outcomes. In accordance with JBI methodology, a risk-of-bias assessment was not performed. RESULTS: A total of 31 eligible studies met the inclusion criteria, examining various LLMs, including OpenAI's GPT series, Gemini, Bard, Claude, Copilot, and Llama. Specifically, GPT-4.0 achieved the most consistent improvements in standardized readability metrics (eg, the Flesch-Kincaid Grade Level [FKGL]). However, achieving predefined target RGLs remained challenging across all LLMs, particularly at lower RGLs. Findings on content fidelity were inconsistent: despite high content similarity scores, content accuracy was often compromised. CONCLUSIONS: This is the first scoping review to comprehensively synthesize evidence on automated technologies for text simplification in PEMs. The review identified 2 critical validation gaps. First, no study examined the linguistic correctness (eg, grammar and typographical errors) of automatically simplified PEMs. Second, and most notably, the understandability of the simplified PEMs was assessed exclusively by experts, with no empirical evaluation involving laypeople. Although LLMs effectively reduce text complexity as measured by objective readability metrics, reliance on these formulas represents a critical limitation, as they serve merely as structural proxies. Improvements in readability do not guarantee the maintenance of content accuracy or laypeople's understandability. Current evidence is further limited by the lack of systematic prompt quality evaluation and the predominant focus on English-language PEMs in US contexts, restricting generalizability. This review provides a foundation for future research by highlighting the need for validated evaluation frameworks that encompass layperson testing and content verification. For clinical practice, LLMs should currently serve as assistive tools, with mandatory expert review remaining essential to verify content fidelity before disseminating LLM-simplified PEMs to laypeople.
12 May 2026
Read appraisal →Scandinavian journal of primary health care
Identifying factors influencing acceptance of artificial intelligence among general practitioners in Danish general practice: a cross-sectional web-based survey study
BACKGROUND: Despite rapid advances in artificial intelligence (AI), its adoption in Danish general practice remains limited and decentralized, relying on individual general practitioners' (GPs') decisions. This study aimed to evaluate Danish GPs' acceptance of AI and the importance of factors derived from established technology-acceptance models. METHODS: A cross-sectional web survey was conducted from mid-September 2024 to March 2025, using a culturally adapted version of a previously validated tool. The survey included 42 items: nine background items and 33 items covering 11 factors, including Medical and Non-Medical Performance Expectancy, Effort Expectancy, Social Influence (Medical and Patient), Facilitating Conditions, Perceived Trust, Anxiety, Professional Identity, Innovativeness and Behavioral Intention. Descriptive statistics, subgroup comparisons and Cronbach's alpha were provided. RESULTS: A total of 109 GPs and residents responded, with 84.4% (n = 92) completing the survey. Attitudes toward AI were generally neutral to positive, with mean scores above neutral in seven of the 11 factors. The most positive factors measured were Medical and Non-Medical Performance Expectancy. Scores related to Behavioral Intention were also high. Lower scores appeared in Perceived Trust, Facilitating Conditions and Anxiety. Older GPs (60 years and above) reported lower scores on Non-Medical Performance Expectancy and Social Influence (Medical); GPs in smaller cities showed more positive attitudes across several factors. Cronbach's alpha indicated good internal consistency on all but two scales. CONCLUSION: In summary, Danish GPs demonstrate a strong intention to adopt AI. However, issues persist as significant barriers regarding Anxiety, Perceived Trust and Facilitating Conditions. Implementation strategies considering context and clinician demographics are recommended.
11 May 2026
Read appraisal →JMIR formative research
Feasibility of Integrating Wearable Devices and Ecological Momentary Assessment for Real-Time Environmental Exposure Estimation: Proof-of-Concept Study
BACKGROUND: Environmental exposures such as heat and air pollution are critical determinants of health, yet traditional assessment methods rely on stationary monitors or residential address proxies that fail to capture the exposures that individuals experience throughout the day. OBJECTIVE: This pilot study aimed to assess the feasibility of integrating ecological momentary assessment (EMA), wearable devices, and continuous GPS tracking to capture real-time environmental exposures and to explore associations with concurrent health outcomes. METHODS: In total, 7 young adults (aged approximately 16 to 24 years; 5/7, 71% female) wore Fitbit Charge 6 watches from July 2025 to August 2025 (mean 28.1, SD 1.1 days), recording sleep quality and duration, resting heart rate, breathing rate, heart rate variability, and physical activity. GPS location measured at up to 5-minute intervals (mean 19.7, SD 25.8 measurements per day) was linked to ambient temperature, humidity, and air pollution data (particulate matter <2.5 um or <10 um in diameter, nitrogen dioxide, sulfur dioxide, ozone, and carbon monoxide) derived from monitoring stations, satellites, and climate models using data-integration algorithms accessed via an application programming interface. EMA surveys administered 3 times per day captured participants' emotional states and location (inside or outside). Feasibility targets were ≥3 GPS measurements per day, ≥1 survey completed per day, and complete sleep data on ≥50% of days. We examined exploratory bivariate correlations between environmental exposures, physiological measures, and self-reported mood, adjusting for multiple comparisons using false discovery rate correction. RESULTS: Of the 7 participants, 5 (71%) met predefined feasibility targets. Mean compliance included 565 (SD 457) GPS coordinates per participant, 1.4 (SD 0.2) EMA surveys per day, and complete Fitbit sleep data on 64% (SD 27%) of days. Surveys identified barriers to compliance, including perceived complexity of the system and forgetting to put the Fitbit watch back on after removing it. Exploratory correlations (6/7, 86% of participants with complete Fitbit data) revealed associations between nitrogen dioxide and heat exposure and reduced heart rate variability (a marker of parasympathetic tone), and between air pollutants (sulfur dioxide) and increased negative emotions. Heat exposure showed a paradoxical pattern of lower self-reported sadness but reduced heart rate variability with higher levels of heat exposure. Given the small sample size, these correlations should be considered preliminary and hypothesis generating rather than definitive findings. CONCLUSIONS: This study demonstrates that the multimodal integration of wearable devices, GPS tracking, and EMA is feasible for capturing real-time environmental exposures and concurrent health outcomes in young adults. This approach addresses critical exposure misclassification issues in environmental health research that relies on residential addresses as proxies. Preliminary patterns suggest complex relationships between environmental exposures and both physiological and emotional outcomes, warranting further investigation in larger, more diverse samples. This approach could inform future personalized environmental health interventions.
11 May 2026
Read appraisal →Journal of medical Internet research
AI in Clinical Decision Support Systems: Promising Applications and Strategies for Managing Data Challenges
The translation of big data analytics and artificial intelligence (AI) into clinical decision support systems (CDSSs) has advanced from proof of concept to real-world clinical practice. AI-informed CDSSs show measurable improvements in diagnostic accuracy, risk stratification, resource use, and patient outcomes compared to traditional models, offering the potential to assist clinicians in managing symptom complexity and uncertainty in health care delivery. Despite this potential, access to large amounts of high-quality and granular data remains one of the most significant bottlenecks to AI-enabled CDSSs. We argue that as health care systems increasingly adopt data-driven decision support, addressing the challenges of data accessibility and protection is essential to realizing the full potential of AI in clinical medicine. We use selected case examples of AI-informed CDSSs in oncology, organ transplantation, diabetic retinopathy, epilepsy, spinal cord injury, rare disease diagnosis, and emergency medicine to illustrate opportunities and challenges related to AI's potential to improve patient outcomes. We discuss public and semipublic, medical institutional and commercial, and government and national data sources that are currently available for the development of CDSSs and highlight the practical and ethical constraints associated with these data. We consider alternative data resources and ways in which health care systems can strengthen data ecosystems to increase AI-driven CDSS efficacy and implementation to improve patient outcomes.
11 May 2026
Read appraisal →JMIR research protocols
Investigating the Implementation and Impact of AI-Assisted Fall Prevention in Hospitals: Protocol for a Multicenter, Multimethod Observational Study in Sweden (SAFE)
BACKGROUND: Artificial intelligence (AI) has the potential to enhance patient safety, particularly in the prevention of in-hospital falls. Recent advances in sensor-based AI systems allow for the analysis of complex, multimodal data to generate real-time alerts, enabling health care professionals to intervene before a fall occurs. By shifting from reactive responses to proactive risk management, these technologies may enable reductions in fall incidence and improvements in care outcomes. As a result, hospitals across Europe are increasingly adopting such systems. Nevertheless, empirical evidence concerning their routine implementation remains limited, particularly concerning their impact on patient safety, clinical workflows, and the usage of health care resources. Addressing these gaps is essential for effective and sustainable integration into hospital care. OBJECTIVE: This paper outlines the protocol for the multicenter, multimethod project SAFE (Safe AI-Assisted Fall Prevention Through Evidence), which investigates the implementation and impact of AI-assisted fall prevention in Swedish hospitals. METHODS: The research project is a collaboration between Halmstad University and hospitals in the Västra Götaland Region (VGR) and will, during 2026-2028, investigate an ongoing large-scale AI system implementation in VGR hospitals, covering up to 2400 patient beds. Using surveys, interviews, observations, and a retrospective study, it will track the implementation and impact over time. Two learning laboratories involving patients, their relatives, and health care professionals will be conducted to codevelop strategies for the implementation of AI-assisted fall prevention. RESULTS: The project will provide evidence-based insights and practical guidance on AI-assisted fall prevention. The findings will be relevant not only to patients, health care professionals, and hospital organizations, but also to policymakers and stakeholders involved in the digital transformation of health care. CONCLUSIONS: Although VGR serves as the primary research setting, the project's results will inform future similar initiatives in Sweden and offer transferable lessons for other health care systems internationally. This study will contribute to the evidence base on AI-assisted fall prevention in health care, supporting the responsible and scalable integration of such systems across diverse health care environments.
11 May 2026
Read appraisal →JMIR formative research
Older Adults' Experiences Navigating Setup of Digital Health Technology: Implementation Report
BACKGROUND: Digital health and connected technologies may support better health outcomes among older adults, including those with multiple chronic conditions or low engagement in health behaviors. However, initial experiences with technology, including during unboxing, setup, and first use, can influence emotional reactions and perceptions and can ultimately determine sustained, meaningful use. Older adults with low technology experience or poor health may be particularly vulnerable to frustration, stress, or abandonment of devices when early interactions are negative. OBJECTIVE: The purpose of this implementation study was to closely observe the initial engagements with a telehealth treatment app and connected blood pressure monitor (BPM) among a group of older adults with low prior technology use and reported low health behavior engagement. The goal was to identify setup "pain points" that may influence initial impressions and intention to use the technology over time. METHODS: A total of 24 older adults (aged ≥65 years) were recruited for a 4-week trial of a telehealth app. Participants were provided with a box containing a tablet preloaded with the app, paper instructions, and a BPM and cuff. Researchers first conducted in-home ethnographic interviews with participants to observe the unboxing and setup process, documenting experiences with reading instructions, using the BPM, and engaging with customer support. Weekly check-in calls and a final exit interview captured ongoing experiences and likelihood of continued use. Interview recordings were transcribed and independently coded, guided by the unified theory of acceptance and use of technology. RESULTS: Most of the sample were White (20/24, 83%) and female (14/24, 58%). Negative experiences with the app's customer support were the top challenge for participants, with representatives providing confusing steps or conflicting terminology. Other common challenges were understanding instructions, connecting to Bluetooth, and correctly using the BPM. While 67% (16/24) of the participants indicated that they were likely or very likely to continue to use the app after the study ended at the end of week 1, this number dropped to 54% (13/24) by the end of the 4 weeks. Participants who reported lower technology self-efficacy at the beginning of the study also experienced frustration, anxiety, and embarrassment as friction with the setup process continued. CONCLUSIONS: First impressions of digital health apps play a critical role in influencing older adults' emotions and perceptions regarding the technology and may impact the likelihood of longer-term engagement. Those with lower technology self-efficacy are particularly susceptible to experiencing negative emotions such as frustration, stress, or shame. Mobile health apps and interventions targeting older adults should incorporate simplified instructions with clear, consistent terminology and well-trained customer support staff to improve the onboarding experience.
11 May 2026
Read appraisal →JMIR formative research
The Role of Trust in Text Messaging for Promoting Patient Portal Activation Among Low-Income Patients: Quality Improvement Project
BACKGROUND: The increasing reliance on patient portals for electronic health records has widened the digital health care access gap, particularly among low-income and Medicaid-insured populations. However, resources exist to assist low-income patients with portal enrollment; in obtaining a free smartphone; and, in New York, in obtaining low-cost internet. Automated bidirectional SMS text messaging offers a scalable and cost-effective strategy for identifying low-income patients' digital health needs and eligibility for resources by using screening questions and providing tailored information on how to access available resources. OBJECTIVE: This study aimed to increase portal access among low-income patients using automated bidirectional SMS text messaging and assess its feasibility and acceptability. METHODS: This quality improvement initiative involved sending automated, bidirectional SMS text messages in English to 12,381 Medicaid-insured and/or low-income patients from a primary care practice. Messages assessed patients' digital health needs and provided adaptive, personalized resources and assistance for enrolling in the patient portal and for accessing digital technology. We assessed response rates and follow-up portal enrollment rates. We surveyed participants regarding the acceptability, appropriateness, and usability of the SMS text messaging intervention, as well as their subsequent use of the patient portal. We performed descriptive statistics and a binomial probability test. RESULTS: In total, 9.2% (1140/12,381) of patients responded to the SMS text messages, with 3.9% (481/12,381) opting out and 5.3% (659/12,381) actively engaging. Among respondents, 71.1% (469/659) completed the follow-up survey. Respondents were predominantly female (336/469, 71.6%), with ages ranging from 18 to 65 years or older. Most respondents rated the message's clarity (420/469, 89.6%), its usefulness (400/469, 85.2%), and the demonstration of care by their health team (350/469, 74.6%) favorably. Concerns regarding privacy (61/469, 13%) and trustworthiness (71/469, 15%) were noted. Notably, 71% of initially unenrolled patients activated their patient portals after the intervention (P=.007), exceeding the hypothesized expectations. CONCLUSIONS: Automated bidirectional SMS text messaging had mixed effects on promoting patient portal use among low-income patients. Response rates to SMS text messages were low when delivered from an unknown phone number. Among responders, most reported that these messages were useful and that they would recommend them to others. Research is needed to determine optimal strategies for introducing the program and vendor phone numbers to patients to improve engagement.
11 May 2026
Read appraisal →PloS one
FairCareNLP: An AI-driven patient review analyzer for healthcare.
OBJECTIVE: To develop and evaluate an automatic patient review analyzer that applies advanced Natural Language Processing (NLP) and machine learning methods to improve the efficiency, fairness, and accuracy of healthcare feedback analysis. MATERIALS AND METHODS: We designed a multi-component pipeline incorporating sentiment analysis, key theme extraction, clinical Named Entity Recognition (NER), and fairness modules. Bias mitigation was addressed through the integration of three complementary approaches: adversarial debiasing, Hard Debiasing, and Iterative Null-space Projection (INLP). Multiple BERT-based models (DistilBERT, BioBERT, RoBERTa-base, BERT-base-uncased) were trained and evaluated under varying hyperparameters and fairness/adversarial loss configurations. Model performance was assessed using accuracy, F1, recall, precision, AUC, Equalized Odds (EOD), and Word Embedding Association Test (WEAT) metrics. RESULTS: Adversarial loss ([Formula: see text]) consistently decreased model performance across accuracy, F1, precision, and recall. In contrast, Hard Debiasing and INLP improved WEAT scores while preserving or enhancing other metrics, with INLP yielding the best overall performance. Specifically, INLP with fairness loss improved EOD by 14%, gender WEAT scores by 15%, and achieved slight gains for ethnicity and socioeconomic WEAT scores. The best model achieved accuracy of 0.856, F1 score of 0.812, recall of 0.798, and precision of 0.829. The key theme analysis module identified 82% of expert-labeled themes, though 21% of patient comments lacked expert labels for valence or related attributes. DISCUSSION: Our results demonstrate the trade-offs between fairness and performance in bias mitigation strategies. While adversarial debiasing reduced predictive accuracy, INLP and Hard Debiasing improved fairness without significant degradation in task performance. Gender bias was easier to mitigate than multi-class sensitive attributes such as ethnicity and income. This difference indicates a need for fairness techniques designed for multi-class sensitive attributes. CONCLUSION: This work presents an NLP pipeline for patient feedback analysis with multiple debiasing strategies. This pipeline improves the fairness and accuracy of insights from unstructured patient reviews and supports inclusive patient-centered care.
11 May 2026
Read appraisal →JMIR human factors
Identifying Skill and Usability Barriers to Digital Health Tool Use Among Older Adult Patients in US Safety Net Clinics: Mixed Methods Study
BACKGROUND: Despite their benefits, digital health tools often face adoption barriers because of the digital divide. Identifying the fundamental user skills required to effectively navigate these tools and the usability barriers is essential to addressing disparities in use. OBJECTIVE: This study aimed to identify the skill and usability barriers to using digital health tools. METHODS: This study included English-, Spanish-, or Cantonese-speaking patients, aged ≥50 years, who received care at an urban safety net health system in the United States. Participants completed a survey examining sociodemographic characteristics and digital health tool use and were observed and video recorded as they navigated four digital health care tasks: (1) launch a video visit, (2) visit a health website through a URL, (3) log in to the patient portal, and (4) sign up for a patient portal account. Participants who could not independently perform the tasks received additional support. Tasks were conducted in English, while instructions and additional assistance were provided in each participant's preferred language. Video recordings were thematically coded to identify the fundamental skills needed for effective digital tool use and usability barriers in the design of digital tools. We examined whether task independence was associated with participant demographics and thematic categories using Kruskal-Wallis, χ2, and Fisher exact tests. RESULTS: In total, 74% (34/46), 52% (31/60), 71% (44/62), and 70% (43/61) of participants (N=64) independently completed digital tasks 1, 2, 3, and 4, respectively. Older age, minoritized races and ethnicities, non-English language preference, lower educational attainment, access to cellular data only or no internet access, and lack of a portal account were associated with a higher likelihood of requiring assistance or being unsuccessful at completing each task (P<.001, except for older age [P=.004]). The qualitative coding of video recordings identified 3, 4, and 6 categories of typing, navigation, and human-computer interaction (HCI) skills, respectively, as fundamental skills required to independently complete digital tasks. χ2 and Fisher exact tests indicated significant associations between most typing, navigation, and HCI categories and independent task completion. We coded usability barriers as one of 6 learnability challenges or 3 operability challenges. CONCLUSIONS: This study identified that independent use of digital health tools requires fundamental typing, navigation, or HCI skills as well as high usability of digital tools. The inclusion of 4 different digital tasks added specificity to the type of skills and usability considerations necessary to ensure accessibility of digital health tools to diverse older adults. This study underscores the need for vendors to cocreate digital health tools with historically excluded end users in mind. As health care systems expand digital tool adoption, they must distinguish fundamental skill gaps from usability barriers, as each may require different intervention strategies.
10 May 2026
Read appraisal →JMIR mHealth and uHealth
mHealth-Based Gamification Interventions to Promote Health Among Older Adults: Scoping Review
BACKGROUND: Healthy aging has emerged as a global priority. However, older adults' participation in health promotion programs remains low, and traditional health promotion models have achieved limited success in fostering sustained engagement among this population. Mobile health (mHealth)-based gamification interventions offer a promising way to address these challenges. However, no published reviews support or oppose the use of mHealth-based gamification interventions as health promotion strategies in older adults. OBJECTIVE: The study aimed to identify mHealth interventions using gamification to promote health among older adults. METHODS: Our scoping review was conducted following the Joanna Briggs Institute recommendations for scoping reviews and Arksey and O'Malley's framework. The process followed PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines and PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Literature Search Extension) checklist. A comprehensive literature search was conducted across 8 databases: PubMed, Scopus, Web of Science, Embase, Cochrane Library, CINAHL, PsycARTICLES, and IEEE Xplore Digital Library, from their inception to December 10, 2025. Two reviewers independently screened titles, abstracts, and full texts via Rayyan, with disagreements resolved by a third reviewer. RESULTS: This scoping review identified 11 studies. Only 1 article was published before 2022. The interventions were found to improve enjoyment and motivation (n=5), cognitive function (n=3), physical activity (n=2), and digital literacy (n=2). Individual studies also reported improvements in mental health (n=1) and adherence (n=1), a reduction in suicidal ideation (n=1), improvements in physical function (n=1), the promotion of social engagement (n=1), and the identification of mild cognitive impairment (n=1). Game elements used were ranked by frequency as progress, challenges, goals, levels, reward, sensation, storytelling or narration, leaderboard, surprise, and avatar. No research was found to use the game element of "social sharing." mHealth types included augmented and virtual reality-based training systems, wearable devices, mobile phones, tablets, and Windows platforms and devices. Notably, only 4 studies applied theoretical frameworks, and 3 omitted the concrete approach to gamification. CONCLUSIONS: As the first scoping review to identify and map mHealth-based gamification interventions for older adults, this study highlights their potential as an innovative approach to health promotion. By systematically synthesizing evidence regarding intervention designs, gamification strategies, and preliminary health outcomes, it establishes a foundation for future inquiry. However, this review is limited by the small number of included studies, precluding broad generalizations. Future research should assess long-term impacts, integrate theoretical frameworks, establish reporting guidelines, design personalized social-interactive interventions, and expand to broader health domains. Ultimately, these insights provide targeted guidance for developing age-appropriate digital health solutions, contributing to the realization of active aging.
10 May 2026
Read appraisal →Current urology reports
Artificial Intelligence in Urologic Documentation: A Review of Emerging Capabilities and the Ongoing Need for Human Oversight
PURPOSE OF REVIEW: Clinical documentation continues to expand in volume and complexity, spanning outpatient encounters, inpatient summaries, patient-portal communications, and educational materials. These growing demands contribute to clinician burden and reduce time available for direct patient care. Artificial intelligence (AI) has emerged as a potential strategy to streamline documentation workflows. This review evaluates current AI applications in clinical documentation, with illustrative examples from urologic practice. RECENT FINDINGS: Ambient AI scribes can capture the bulk of outpatient encounters and generate structured draft notes that clinicians edit rather than write de novo. Large language models have shown promise in assisting with inpatient documentation and discharge summaries, often producing drafts that are coherent, readable, and shorter than physician-authored text. AI tools can also simplify patient education materials and translate dense radiology reports into accessible language. Across these domains, however, studies consistently demonstrate that AI-generated content remains vulnerable to factual errors, omissions, hallucinations, and misaligned emphasis, reinforcing the need for clinician oversight. Overall, emerging evidence supports a complementary relationship between clinicians and AI. Used as supervised drafting aids rather than autonomous authors, AI tools have the potential to ease documentation burden and create more time for direct patient care without diminishing the clinician's role in shaping the medical record.
10 May 2026
Read appraisal →Diagnostic and interventional radiology (Ankara, Turkey)
Foundation models for radiology: fundamentals, applications, opportunities, challenges, risks, and prospects.
Foundation models (FMs) represent a significant evolution in artificial intelligence (AI), impacting diverse fields. Within radiology, this evolution offers greater adaptability, multimodal integration, and improved generalizability compared with traditional narrow AI. Utilizing large-scale pre-training and efficient fine-tuning, FMs can support diverse applications, including image interpretation, report generation, integrative diagnostics combining imaging with clinical/laboratory data, and synthetic data creation, holding significant promise for advancements in precision medicine. However, clinical translation of FMs faces several substantial challenges. Key concerns include the inherent opacity of model decision-making processes, environmental and social sustainability issues, risks to data privacy, complex ethical considerations, such as bias and fairness, and navigating the uncertainty of regulatory frameworks. Moreover, rigorous validation is essential to address inherent stochasticity and the risk of hallucination. This international collaborative effort provides a comprehensive overview of the fundamentals, applications, opportunities, challenges, and prospects of FMs, aiming to guide their responsible and effective adoption in radiology and healthcare.
10 May 2026
Read appraisal →European journal of pediatrics
Metaphorical perceptions of secondary school students regarding the concept of generative artificial intelligence
UNLABELLED: The rapid integration of generative artificial intelligence (GenAI) tools, such as ChatGPT, into educational contexts has raised important questions regarding how adolescents conceptualize and make sense of these technologies. Understanding students' perceptions is essential for developing age-appropriate, ethical, and pedagogically sound approaches to AI use in secondary education. This descriptive qualitative study employed a phenomenological approach and metaphor analysis to explore secondary school students' perceptions of generative artificial intelligence. The study sample consisted of 332 students aged 14-18 years from four secondary schools in Türkiye. Data were collected using an open-ended prompt ("Generative artificial intelligence is like … because …") and analyzed through content analysis. Metaphors were categorized based on shared semantic and conceptual features, and inter-rater reliability was established using Cohen's kappa (κ = 0.92). Analysis revealed ten metaphor categories clustered under five overarching themes: generative artificial intelligence as (1) a source of knowledge, (2) a teaching and guiding entity, (3) a supportive and assisting tool, (4) a reflection of human intelligence, and (5) a dual-purpose (beneficial-risky) technology. Students most frequently conceptualized GenAI as a comprehensive knowledge source (e.g., book, encyclopedia) and as a human-like cognitive entity (e.g., brain, wise person). At the same time, metaphors reflecting ethical awareness and potential risks, such as misuse and overreliance, were also identified. The findings indicate that secondary school students hold multifaceted and nuanced perceptions of generative artificial intelligence, encompassing both educational opportunities and ethical concerns. These results highlight the importance of integrating AI literacy into secondary education in ways that promote critical thinking, responsible use, and awareness of GenAI's limitations alongside its potential benefits. CONCLUSION: It was determined that secondary school students perceive generative artificial intelligence ambivalently as both a useful tool and a source of ethical and emotional concern, highlighting the need for developmentally appropriate artificial intelligence literacy approaches. WHAT IS KNOWN: • GenAI tools such as ChatGPT are increasingly integrated into educational contexts and have the potential to support personalized learning, information access, and student engagement. • Existing research has primarily focused on educators' perspectives or higher education settings, while studies examining adolescents' perceptions of GenAI remain limited. WHAT IS NEW: • This study provides empirical evidence on secondary school students' metaphorical perceptions of generative artificial intelligence within a K-12 context. • Findings reveal that adolescents conceptualize GenAI in multifaceted ways, including as a knowledge source, teaching and guiding entity, supportive tool, reflection of human intelligence, and a dual-purpose (beneficial-risky) technology.
10 May 2026
Read appraisal →Annals of emergency medicine
Ambient Artificial Intelligence Scribe Adoption and Documentation Time in the Emergency Department
STUDY OBJECTIVES: To describe real-world adoption of an ambient artificial intelligence (AI) scribe in the emergency department (ED) and compare documentation time and note characteristics between ambient and standard encounters using electronic health record audit logs. METHODS: We performed a retrospective observational study of adult ED encounters at a tertiary academic medical center. Attending physicians could optionally use an ambient AI scribe to generate notes from patient-clinician conversations. We included single-attending encounters in core ED zones and excluded visits with human scribes. Electronic health record audit logs provided documentation of time during and after the shift, total electronic health record time, and note length. We summarized adoption by physician, zone, and acuity and compared medians between ambient and standard encounters. RESULTS: Among 8,740 eligible encounters, 976 (11.2%) used ambient AI. Thirty-five of 92 attendings (38%) used the tool, and a small group of high-frequency users accounted for most ambient encounters. Ambient use clustered in telemedicine and vertical-care zones (chair-based ambulatory care) and in lower-acuity patients, as well as those not requiring interpreters. Median on-shift documentation time was 2:45 min for ambient encounters versus 3:50 min for standard encounters (difference -1:05; -28%). Median total electronic health record time was 8:39 min versus 10:21 min (-16%), and ambient notes were shorter overall. CONCLUSION: Early ED implementation of ambient AI scribes showed low but highly skewed adoption, with physicians favoring lower acuity, noninterpreted encounters. When used, ambient AI was associated with shorter on-shift documentation time, total electronic health record time, and note length.
10 May 2026
Read appraisal →The British journal of psychiatry : the journal of mental science
Characterisation of serious mental illness trajectories through transdiagnostic clinical features
BACKGROUND: Electronic health records (EHRs), increasingly available in low- and middle-income countries (LMICs), provide an opportunity to study transdiagnostic features of serious mental illness (SMI) and its trajectories. AIMS: Characterise transdiagnostic features and diagnostic trajectories of SMI using an EHR database in an LMIC institution. METHOD: We conducted a retrospective cohort study using EHRs from 2005-2022 at Clínica San Juan de Dios Manizales, a specialised mental health facility in Colombia, including 22 447 patients with schizophrenia (SCZ), bipolar disorder (BPD) or severe/recurrent major depressive disorder (MDD). Using diagnostic codes and clinical notes, we analysed the frequency of suicidality and psychosis across diagnoses, patterns of diagnostic switching and the accumulation of comorbidities. Mixed-effect logistic regression was used to identify factors influencing diagnostic stability. RESULTS: High frequencies of suicidality and psychosis were observed across diagnoses of SCZ, BPD and MDD. Most patients (64%) received multiple diagnoses over time, including switches between primary SMI diagnoses (19%), diagnostic comorbidities (30%) or both (15%). Predictors of diagnostic switching included mentions of delusions (odds ratio = 1.47, 95% CI 1.34-1.61), prior diagnostic switching (odds ratio = 4.01, 95% CI 3.7-4.34) and time in treatment, independent of age (log of visit number; odds ratio = 0.57, 95% CI 0.54-0.61). Over 80% of patients reached diagnostic stability within 6 years of their first record. CONCLUSIONS: Integrating structured and unstructured EHR data reveals transdiagnostic patterns in SMI and predictors of disease trajectories, highlighting the potential of EHR-based tools for research and precision psychiatry in LMICs.
9 May 2026
Read appraisal →Journal of the American Medical Informatics Association : JAMIA
Evaluating the potential of fast healthcare interoperability resources for clinical registry data submission.
OBJECTIVE: We conducted the Clinical Registry Extraction and Data Submission (CREDS) project to evaluate the readiness of HL7 Fast Healthcare Interoperability Resources (FHIR) for provisioning data from health information systems for the American College of Cardiology Cardiac Catheterization Percutaneous Coronary Intervention (CathPCI) Registry. MATERIALS AND METHODS: The CREDS project had 3 workstreams: (1) evaluation of the readiness of clinical documentation for data transforms, (2) modeling of a FHIR-based clinical workflow for registry data submission, and (3) development and demonstration of a CREDS FHIR implementation for registry data submission. RESULTS: Of the 344 data concepts comprising the CathPCI Registry, only 111 (32%) were sufficiently discrete to be listed in the CathPCI Data Dictionary with a terminology mapping. Cardiologist informaticians identified an additional 42 concepts suitable for provisioning via a FHIR payload. The resulting notional workflow combined FHIR-based data assembly with manual chart abstraction of compound, summative, and complex clinical concepts. A CathPCI FHIR StructureDefinition artifact was authored, incorporated into a CREDS FHIR Implementation Guide, and balloted to Standard for Trial Use status. DISCUSSION: CREDS demonstrated both potential and limitations for using FHIR for registry data submission. The largest technical impediment was the volume of code (>11 000 lines) for the FHIR StructureDefinition. Lack of regularized clinical vocabularies, reliance of registries on complex clinical concepts, and absence of FHIR infrastructure must be overcome before CREDS can be used at scale. CONCLUSION: CREDS demonstrated proof-of-concept FHIR-based provisioning of clinical data for registry submission. All artifacts are open source to inform others with similar interests.
9 May 2026
Read appraisal →Journal of medical systems
Interpretable Machine Learning to Anticipate the Diagnostic Yield of EEG in the Emergency department. The EMINENCE study.
INTRODUCTION: Emergent electroencephalography (emEEG) is increasingly employed in the emergency department (ED) for evaluating altered consciousness and seizure-related conditions, yet standardized criteria guiding its use remain limited. METHODS: We retrospectively analyzed 1,018 patients (mean age 66 ± 20 years; 48.4% female) undergoing emEEG at the ED of the Careggi Teaching Hospital (Florence, Italy) in 2023. Clinical, anamnestic, and neuroimaging data available at admission were used to train supervised machine-learning (ML) models. We evaluated tree-based ensembles (Random Forest and XGBoost) to predict abnormal and epileptiform emEEG, as well as confirmation or refutation of initial diagnosis. Ground-truth labels were derived from a multidisciplinary expert team including neurologists, neurophysiopathologists and intensivists. Model performance was assessed with 5 × 5 nested cross-validation, receiver operating characteristic (ROC) analysis, balanced accuracy, decision-curve analysis, and Shapley Additive Explanations (SHAP) interpretability. RESULTS: Abnormal emEEG occurred in 691 cases (67.9%), epileptiform activity in 192 patients (18.9%). emEEG ruled out the initial diagnostic suspicion in 514 cases (50.5%) and confirmed it in 188 cases (18.5%). Best performance was obtained with Random Forest for abnormal emEEG (AUC 0.79, 95% CI: 0.76-0.82) and diagnosis rule-out (0.84, 0.81-0.86), and with XGBoost for epileptiform emEEG (0.82, 0.78-0.85) and diagnosis confirmation (0.82, 0.79-0.85). Performance varied by initial diagnostic suspicion, but subgroup-stratified analyses showed overall consistent patterns. Key predictive features included altered consciousness, prior brain lesions, antiseizure therapy, and seizure-related presentations. Interpretability analyses revealed seizure-centric features drove confirmation, while systemic or nonspecific features favored refutation. CONCLUSIONS: Interpretable ML models using only admission data can predict emEEG outcomes and anticipate their diagnostic contribution, supporting triage and decision-making in emergency neurology without replacing clinical judgment. Models and explanations were easily usable on a freely-accessible website ( www.emergencyeeg.com ), where tools return probabilistic outputs for all four prediction tasks together with per-patient explanation plots, enabling transparent and reproducible use. CLINICAL TRIAL NUMBER: Not applicable.
9 May 2026
Read appraisal →International journal of nursing studies
The potential of digital health technologies in saving nursing resources: A scoping review
BACKGROUND: The global nursing shortage crisis presents a significant challenge to healthcare systems. Digital health technologies, such as communication tools, automation systems, monitoring devices, and information platforms have been proposed as one solution to alleviate the issue by optimizing nursing resources. However, a comprehensive overview of the use and potential of these technologies in optimizing nurses' work and resources is still lacking. OBJECTIVES: The aim of this review is to provide an overview of (a) the digital health technologies used in nursing that may have potential to save nursing resources, (b) which indicators are used to measure the effectiveness of these technologies, and (c) which technologies are found to be effective in terms of saving nursing resources. METHODS: We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for scoping reviews in accordance with the Joanna Briggs Institute (JBI) Manual for Evidence Synthesis. The databases PubMed, CINAHL, and Web of Science were searched. Studies were included if they were peer-reviewed, quantitative, addressed nursing professionals, and considered digital health technologies that specifically aimed to save nursing resources. Relevant data were extracted and synthesized narratively, using an emerging framework. RESULTS: A total of 115 studies were included in this review. Digital health technologies fell into four categories: communication, automation, monitoring, and information technologies. All showed potential for saving resources, with monitoring technologies most often reported as having high potential. Indicators used to measure effectiveness focused on temporal, workforce, and workload resources, and related outcomes including quality of care, patient safety, and cost efficiency. Workforce resources were the most frequently optimized, followed by improvements in patient safety and quality of care. While positive effects were predominant, some studies reported adverse and insignificant effects, indicating variability in effectiveness. CONCLUSIONS: Digital health technologies offer promising opportunities to alleviate nursing resource shortages, but their potential seem to vary by type. Monitoring technologies showed the most consistent benefits, while communication and information technologies had mixed effects and automation technologies require further research to clarify their potential. Given the wide range of indicators used to measure potential resource savings, defining common, standardized indicators is essential for systematically assessing the impact of digital health technologies on nursing work and resources. This would enable comparability across studies, strengthen evidence-based decision-making and guide implementation strategies.
9 May 2026
Read appraisal →Neural networks : the official journal of the International Neural Network Society
Alleviating noise memorization for adversarially robust few-shot learning
Few-Shot Learning (FSL) enables models to learn from just a few examples of new classes by leveraging knowledge from base classes. While FSL has made significant strides, its vulnerability to adversarial attacks-especially with limited data-has been overlooked. To address this, adversarial training is often used to build more robust models. However, we found that this approach can lead the model to memorize adversarial noise, which harms its ability to generalize. Hard labels exacerbate this issue by pushing the model toward perfect accuracy on adversarial examples, while also making it less robust to small changes in weights. To solve these problems, we propose Alleviation of Noise Memorization (ANM), a method that includes Adaptive Label Smoothing for more flexible supervision and Robust Weight Learning to enhance model stability. Our extensive experiments show that ANM effectively reduces noise memorization and improves generalization, outperforming current benchmarks.
6 May 2026
Read appraisal →International journal of medical informatics
Harmonizing patient-reported outcome measures for nasal complaints using traditional and machine learning methods
BACKGROUND: Nasal obstruction measurement instruments are widely used in the field of nasal surgery. There are various scales that measure nasal obstruction and they differ regarding the number of items, their wording, and the type of response options. In order to pool the data and analyze it together, it is necessary to harmonize it so that we can compare participants' nasal obstruction scores irrespective of instrument they filled in. Data harmonization is still not used in the field of nasal obstruction assessment. THE AIM: The aim of this study was to find the best harmonization method in terms of predicting the scores on a target instrument based on the scores from another instrument as precise as possible in the case of four different nasal complaints instruments. A method was sought to find a transformation of scores on the NOSE, Utrecht-Q and SCHNOS that makes them equivalent to ENFAS scores. METHODS: A total of 1324 unique patients completed all four measurement instruments. We tried linear equating, Item Response Theory (IRT), and the following machine learning methods: linear regression, random forest regression, support vector machine regression, and neural network. We used the root-mean-square error (RMSE) of differences between predicted and observed scores to evaluate the quality of harmonization in 5-fold cross-validation. RESULTS: The ML methods gave overall the best results (the lowest RMSEs) and outperformed IRT (which is considered as a common choice for data harmonization in psychometrics). CONCLUSION: The ML methods led to the best quality of the results, confirming their strong potential for data harmonization. This study shows that next to linear equating and IRT that are commonly used for data harmonization, we can also use ML methods for the same purpose and, by doing so, to even increase the quality of the harmonization in certain use cases.
6 May 2026
Read appraisal →Healthcare management forum
Navigating Privacy in Health Data Sharing: A Patient-Centric Approach to Health Information Exchange
Personal Health Information (PHI) sharing through Health Information Exchange (HIE) enhances patient safety in Canada, yet not all provinces and territories voluntarily disclose PHI on safety incidents to federal and pan-Canadian surveillance systems. A frequently cited barrier by healthcare organizations for HIE between different interoperable health databases is patients' and families' concerns for their privacy. This explorative qualitative study examined patients' and families' attitudes toward PHI sharing, including its secondary use in patient safety events. Rather than expressing reluctance, participants described support for HIE when privacy safeguards, such as defined sharing purposes and anonymous formats, were in place. These findings present a significant opportunity for health leaders and data custodians to use the research findings to create a patient-centric framework for the HIE of PHI.
6 May 2026
Read appraisal →Annals of epidemiology
Population-based clustering of co-occurring social determinants: An application of unsupervised machine learning
PURPOSE: This study aimed to develop a cluster-based measure of multiple co-occurring social determinants of health by applying unsupervised machine learning to a population-based cohort, offering a data-driven approach to organize complex social exposures. METHODS: Unsupervised clustering was applied to a population-based cohort of Ontario respondents to six-cycles of the Canadian Community Health Survey (2001-2012) linked to the Canadian census and vital statistics data. Clusters were evaluated using internal metrics, visualization techniques, descriptive analysis and theoretical considerations to determine the optimal number of clusters. Sensitivity analyses were integrated across the iterative clustering process. Premature mortality rates were generated assess validity. RESULTS: Optimal clustering solutions included 4-clusters and 6-clusters. Both cluster solutions revealed distinct social typologies. The 6-cluster solution offered greater granularity and theoretical interpretability. The 4-cluster solution showed greater heterogeneity within certain marginalized groups. Premature mortality rates differed meaningfully across clusters, supporting the clustering approach in capturing risk associated with social exposure. CONCLUSIONS: Unsupervised machine learning methods identified meaningful population subgroups reflecting complex patterns of social exposures. This approach offers a flexible, data-driven method for characterizing social exposures that can be considered alongside theoretical frameworks and used for equity monitoring, intervention planning and policy development.
6 May 2026
Read appraisal →Journal of clinical ultrasound : JCU
Illuminating Research Dynamics: Medical Ultrasound and Deep Learning
BACKGROUND: Deep learning (DL) has enabled advances in ultrasound imaging, but challenges like limited datasets and device variability hinder progress. This study provides the first bibliometric overview of DL research in medical ultrasound. METHOD: We retrieved related publications (2004-Apr 2025) on medical ultrasound and DL from the Web of Science Core Collection. Bibliometric tools (Bibliometrix, VOSviewer, CiteSpace) were used to quantify publication growth, country/institution contributions, and thematic evolution. RESULTS: A total of 3386 publications were included, with the annual output increasing steadily at an average growth rate of 31.02%. While China led all countries in publication volume (1355), the United Kingdom achieved the highest citation rate per article (55.5 citations). Fudan University was the most prolific institution (110), and key authors such as Luca Saba and Jasjit S. Suri formed the principal collaboration hubs. In terms of venues, Diagnostics published the most papers, whereas IEEE Transactions on Medical Imaging exerted the greatest citation impact (5646 citations; IF 8.9). Keyword analysis revealed core themes in classification, segmentation, and CNN-based diagnosis. From 2020 onward, ultrasound-specific topics (e.g., diagnosis, images) dominate; advanced architectures (e.g., U-Net) and emerging clinical concepts such as noninvasive assessment. CONCLUSION: This mapping highlights a rapidly growing, internationally collaborative field. Emerging trends include advanced artificial intelligence (AI) architectures (U-Net, attention-based networks, vision Transformers) and a shift toward surgical and therapeutic applications. These insights underscore the need for multicenter validation, standardized data protocols, and clinically interpretable AI models to accelerate safe translation into practice.
5 May 2026
Read appraisal →The Journal of nursing education
Enhancing Nursing Students' Interdisciplinary Competence via ARCS-Based Digital Health Care Design
BACKGROUND: Digital literacy is increasingly essential in modern nursing education, yet many students lack confidence and prior exposure to digital tools. METHOD: A 6-week project-based course, guided by the ARCS motivational model comprising Attention, Relevance, Confidence, and Satisfaction, engaged 40 first-year nursing students in designing health care applications using the Figma digital platform. The instructional process was organized into three progressive phases: motivational induction, design implementation, and final presentation. RESULTS: Students demonstrated a significant increase in learning motivation (t = 15.784, p < .001) and rated Figma's usability as excellent (System Usability Scale = 84.87). Reflections indicated enhanced digital confidence, interdisciplinary collaboration, and the ability to translate nursing knowledge into practical applications. CONCLUSION: Integrating motivational strategies with project-based learning and an intuitive digital platform effectively promoted student engagement, skill acquisition, and meaningful learning. This model offers a scalable approach for embedding digital competence into nursing curricula.
5 May 2026
Read appraisal →Journal of medical systems
Automated data extraction from electronic medical records for pragmatic clinical trials
Data collection in randomized trials is expensive and labor intensive. With the rise in ongoing pragmatic trials, the use of electronic medical records (EMR) as a source of data has increased. Although potentially faster and cheaper, EMR use can lead to errors. Therefore, to ensure accurate data collection and to avoid systematic errors we performed a study comparing automated data extraction (ADE) with manual data extraction (MDE). We performed a retrospective cohort study to compare the accuracy of ADE using Structured Query Language with MDE by blinded physicians from our EMR. We tested the interrater agreement and intraclass correlation coefficient of clinical baseline data and outcomes of a random sample of 30 patients admitted to the ICU, on mechanical ventilation, requiring opioids for analgosedation for an upcoming pragmatic clinical trial. Key data compared included, but not limited to, patient's demographics, laboratory and vital signs, daily morphine milligram equivalent (MME), days alive and free of mechanical ventilation, days alive and free of hospitalization, days alive and free of ICU, days alive and free of vasopressors, and death. Among 238 patients screened over 1-month period, 72 fulfilled inclusion criteria and 30 were randomly selected to be included in the evaluation. We blindly collected 1320 baseline data, 2160 categorical outcomes and 705 continuous outcomes for a total of 4185 data points. The intraclass correlation coefficient and the Cohen's Kappa were perfect or almost perfect for all data, including outcomes such as daily MME, days alive and free of mechanical ventilation, days alive and free of ICU and days alive and free of hospital with p < 0.001. Among all rechecked data, the ADE was correct in 53 (77.9%) of cases, while MDE in 15 (22.1%). The inaccurate data collected by ADE accounted for 0.36% of the total data-points. The performance of ADE had almost perfect agreement for all outcomes and when rechecking for disagreements, it was more accurate than MDE.
5 May 2026
Read appraisal →Journal of medical imaging and radiation sciences
Explainability-informed benchmarking of two deep learning models for organ-at-risk segmentation in MR-guided adaptive radiotherapy
INTRODUCTION/BACKGROUND: Segmentation of gastrointestinal (GI) organs-at-risk (OARs) is a critical yet time-consuming step in MR-guided adaptive radiotherapy (MRgRT), with manual delineation prone to inter- and intra-observer variability. While deep learning approaches have shown promise, their clinical adoption requires not only accuracy but also interpretability and reliability. This study benchmarks two widely used convolutional architectures, U-Net and Residual U-Net (ResUNet), for abdominal OAR segmentation, with an emphasis on explainability-oriented quantitative analysis. METHODS: An anonymized abdominal MRI dataset was used to train and evaluate U-Net and ResUNet using a 5-fold stratified group cross-validation strategy. Segmentation performance was assessed using the Dice Similarity Coefficient (DSC), Intersection-over-Union (IoU), and the 95th percentile Hausdorff Distance (HD95). Explainability was investigated using Gradient-weighted Class Activation Mapping (Grad-CAM) computed from the final convolutional layer of each network. To enable objective analysis beyond qualitative visualization, Grad-CAM activation maps were quantified using numerical localization metrics relative to ground-truth organ masks, including in-organ energy ratio, boundary energy ratio, pointing accuracy, activation Dice coefficient, centroid distance and activation entropy. Grad-CAM metrics were aggregated across gastrointestinal organs and averaged over the five validation folds. RESULTS: Both architectures demonstrated comparable segmentation performance across organs, with no statistically significant differences across evaluated metrics. Grad-CAM analysis showed similar region-level attention patterns, with in-organ activation ratios of 71.4 ± 8.6% for U-Net and 66.2 ± 9.1% for ResUNet, boundary energy ratios of 24.1 ± 4.9% and 21.8 ± 5.2%, respectively, and pointing accuracies exceeding 70% for both models. Uncertainty analysis based on inter-fold variability and boundary error dispersion indicated comparable stability and bounded worst-case behavior. DISCUSSION/CONCLUSION: By integrating performance, uncertainty and explainability quantitative indicators, this study provides an informed benchmarking of two deep learning models for abdominal OAR segmentation. The results suggest that both U-Net and ResUNet exhibit stable and interpretable behavior under the evaluated configurations, supporting their potential use in MR-guided adaptive radiotherapy workflows where reliability and clinical trust are essential.
4 May 2026
Read appraisal →Journal of medical imaging and radiation sciences
Automatic field-of-view planning for magnetic resonance shoulder imaging using Deep Learning
INTRODUCTION: Accurate prescription of oblique coronal and oblique sagittal field of views (FOV) is essential for diagnostic shoulder MRI. Manual planning is radiographer-dependent, time-consuming, and subject to inter- and intra-operator variability, leading to inconsistent image quality and incomplete coverage. Although deep learning (DL) has advanced automated scan planning in non-oblique planes, oblique shoulder prescriptions remain underexplored; an automated DL approach could standardize FOV prescription, reduce operator dependence, and improve reproducibility and workflow without compromising diagnostic quality. METHODS: In this retrospective multicenter study, 575 shoulder MRI examinations (2019-2025) from four sites were included. Sites A (n=151) and B (n=220) were used for training; testing was performed on sites C (n=61), and D (n=143). A two-stage pipeline was implemented using five oriented bounding box (OBB) variants of YOLOv11 (n, s, m, l, x): Stage 1 performed slice selection; Stage 2 performed FOV prescription. Performance was evaluated against radiographers' prescriptions using mean absolute slice difference (MASD, slices), intersection over union (IoU), and mean absolute angle difference (MAAD, degrees). Clinical utility was assessed by three raters. RESULTS: The YOLOv11-OBB-l model achieved the lowest MASD for Stage 1 (1.016±0.153 slices). For Stage 2, YOLOv11-OBB-x performed best (coronal IoU, 0.847±0.003; sagittal IoU, 0.852±0.007; MAAD, 3.259±0.190°). During testing across each site, MASD ranged from 0.700±0.837 to 1.192±2.550 slices; MAAD from 2.811±2.348 to 4.396±7.158°; coronal IoU from 0.800±0.092 to 0.872±0.065; and sagittal IoU from 0.824±0.111 to 0.887±0.047. Mean clinical utility was 97.2%. Performance was noninferior to interrater variability across all sites and metrics. CONCLUSION: DL-based automated FOV prescription for shoulder MRI achieves performance comparable to radiographers, generalizes across institutions, and demonstrates high clinical utility.
4 May 2026
Read appraisal →Social science & medicine (1982)
Imaginaries of digital health: Unpacking the platformization of mental health through users' practices
The platformization of healthcare is reshaping the organization of mental health services, redefining access, relationships, and meanings of care. This paper examines how users experience and negotiate the platformization of mental health through the case of Unobravo, an Italian digital platform offering online psychotherapy. Drawing on Science and Technology Studies and digital health literature, the study explores how platform imaginaries of efficiency, accessibility, and personalization are enacted and contested in everyday practices. Adopting a mixed-methods design, the research combines qualitative interviews (N = 12) with a quantitative survey (N = 3870) to analyse users' narratives and patterns of engagement. Findings reveal a persistent tension between the promise of algorithmic efficiency and the realities of situated practice. While users value the flexibility and accessibility provided by digital infrastructures, they also encounter infrastructural misalignments, such as unstable connections and opaque algorithmic matching processes. These challenges are addressed through practices of domestication that reconfigure the therapeutic setting, integrating care into mobile and domestic environments. By analysing user practices, the paper investigates how mental health is domesticated through digital platforms and shaped by platform imaginaries, reconfiguring the therapeutic setting as a datafied and everyday space.
4 May 2026
Read appraisal →Clinical pharmacology and therapeutics
Off-the-Shelf Large Language Models for Guiding Pharmacoepidemiological Study Design
This study aimed to assess the ability of two off-the-shelf large language models, ChatGPT and Gemini, to support the design of pharmacoepidemiological studies. We assessed 48 study protocols of pharmacoepidemiological studies published between 2018 and 2024, covering various study types, including disease epidemiology, drug utilization, safety, and effectiveness. The coherence (i.e., "Is the response coherent and well-formed, or is it difficult to understand?") and relevance (i.e., "Is the response relevant and informative, or is it lacking in substance?") of the large language models' responses were evaluated by human experts across seven key study design components. Coding accuracy was assessed. Both large language models demonstrated high coherence, with over 90% of study components rated as "Strongly agree" by experts for most categories. ChatGPT achieved the highest coherence for "Index date" (97.9%) and "Study design" (95.8%). Gemini excelled in "Study outcome" (93.9%) and "Study exposure" (95.9%). Relevance, however, was more variable, with ChatGPT aligning with expert recommendations in over 90% of cases for "Index date" and "Study design" but showing lower agreement for covariates (65%) and follow-up (70%). Coding agreement percentages reveal varying levels of concordance, with the Anatomical Therapeutic Chemical classification system coding system demonstrating the highest agreement at 50% with experts. In contrast, the Current Procedural Terminology and International Classification of Diseases systems showed agreements of 22.2% and 20%, respectively. While ChatGPT and Gemini show promise in certain tasks supporting pharmacoepidemiological study design, their limitations in relevance and coding accuracy highlight the need for critical oversight by domain experts.
4 May 2026
Read appraisal →JMIR medical education
Health Professional Students' Use of Generative Artificial Intelligence During Clinical Placements: Cross-Sectional Online Survey Study
BACKGROUND: Generative artificial intelligence (GenAI) is rapidly expanding in higher education and clinical practice. However, its use during clinical placements, where cognitive demands and responsibility for patient care increase, remains insufficiently documented. OBJECTIVE: This study aimed to characterize self-reported GenAI use during clinical placements, perceived benefits and risks, and related training and governance needs. METHODS: We conducted a cross-sectional online survey at a French university (July 17 to September 30, 2025). Eligible participants were students in medicine, pharmacy, nursing, midwifery, or physiotherapy who were currently in, or had completed within the past 18 months, a clinical placement. A 61-item questionnaire (comprising closed- and open-ended items) assessed GenAI use, task patterns, perceived benefits or risks, and training or governance needs. A composite index classified self-perceived GenAI maturity as minimal, limited, moderate, or high. Group comparisons used χ2 tests; maturity gradients used trend tests. RESULTS: A total of 388 students responded (n=308, 79.4% women), mainly nursing students (n=217, 55.9%). Overall, 204 (52.6%) students reported using GenAI during clinical placements. Use differed across disciplines (χ24=10.71; P=.03), with lower uptake in midwifery (6/23, 26%; odds ratio 0.30, 95% CI 0.11-0.77). Adoption increased markedly with self-perceived maturity (minimal: 2/22, 9% vs high: 22/29, 76%; trend P<.001). Among the 204 users, the most commonly reported uses were information retrieval (n=159, 77.9%), bibliographic search (n=152, 74.5%), and translation or rephrasing (n=145, 71.1%); patient-facing activities were less frequently reported (eg, patient-document drafting or communication preparation: n=78, 38.2%). Although most users reported never entering direct patient identifiers, 48 (23.5%) reported at least 1 disclosure of patient-identifying information, and 96 (47.1%) reported processing real medical content perceived as anonymized. The most endorsed perceived benefits among the 388 students were documentation support (n=315, 81.2%) and improved access to information (n=266, 68.5%). The most endorsed risks were dependency (n=353, 90.9%), skill erosion (n=329, 84.8%), and confidentiality breaches (n=339, 87.4%). Training needs were highest for ethics or regulatory training (294/378, 77.7%) and a best-practice clinical guide (292/373, 78.3%). CONCLUSIONS: GenAI is already used by a substantial proportion of French students in health professions during clinical placements, predominantly for information and documentation support rather than patient-facing activities. Self-perceived readiness is strongly associated with adoption. Reported disclosures and concurrent concerns about dependency, skill erosion, and confidentiality support the need for structured curricula and clear governance frameworks to enable responsible, patient-centered integration of GenAI into clinical education.
3 May 2026
Read appraisal →JMIR medical informatics
Leveraging MedlinePlus to Improve Health Information Access Among Patients and Caregivers: Systematic Literature Review
BACKGROUND: MedlinePlus, developed by the National Library of Medicine (NLM) in the United States, is one of the most widely used, authoritative, consumer-grade health information resources on the web. Although extensively used and discussed in scholarly work for health literacy and patient education, it is unclear how MedlinePlus has been integrated into clinical care or embedded within health informatics applications. OBJECTIVE: This study aimed to understand how MedlinePlus has supported patients and caregivers by increasing access to health information for clinical care and illness management. The insights on this topic will inform the design and development of patient-facing digital health intervention tools for improved health communication, decision engagement, informed decision-making, and health outcomes. METHODS: We conducted a systematic literature review following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. First, we developed a comprehensive literature search strategy, searched 9 citation databases, and aggregated and deduplicated search results before importing them into Covidence for manual screening using predefined inclusion and exclusion criteria. Second, reviewers independently assessed all studies at the title-abstract and full-text levels, resolving discrepancies through ongoing discussions. Third, we applied the PICO (problem/population, intervention, comparison, and outcome) and the Collaborative Chronic Care Model as guiding frameworks for data extraction and analysis. All included studies underwent quality assessment using the Mixed Methods Appraisal Tool. RESULTS: In total, 28 studies reported in 27 sources met our inclusion criteria. We categorized the extracted data into 4 areas. First, regarding bibliometrics, the studies were reported between 2004 and 2024, with 2010 having the highest number of studies. Of these studies, 25 were conducted in the United States, 2 were conducted in Iran, and 1 was conducted in Argentina. Health informatics journals and conference proceedings, as well as library science journals, were prominent publishing venues. The NLM funded half of the studies. Second, regarding participants, most studies focused on outpatients. Other participant roles included physicians, nurses, hospital staff, pharmacists, and librarians. Fewer than half of the studies addressed the social determinants of health. Third, regarding intervention, most studies implemented MedlinePlus information interventions within clinical settings. Other interventions occurred in community pharmacies, community organizations, libraries, online health platforms, or patient portals. Fourth, regarding outcome, only 4 studies assessed clinical outcomes, and the findings were mixed and inconsistent. However, 24 of 28 studies reported positive nonclinical outcomes, including improved attitudes toward and satisfaction with MedlinePlus and enhancements in patients' information-seeking behaviors, confidence, and willingness to engage in decision-making, physician-patient communication, self-management, and self-efficacy. CONCLUSIONS: This systematic literature review is the first comprehensive examination of how MedlinePlus has been integrated into clinical care, supporting patients and caregivers with enhanced access to health information. Our findings offer evidence and insights through the Collaborative Chronic Care Model lens and can guide the development of digital health interventions to improve patient health.
3 May 2026
Read appraisal →PloS one
Enhanced graph attention network by integrating Long Short-Term Memory for artificial emotion representation in multi-modality datasets
Emotion representation is a critical aspect of artificial intelligence, particularly in human-computer interaction and affective computing. Emotion recognition from multi-modal data remains challenging due to the complex semantic relationships between textual, audio, and visual features. This study proposes a hybrid model combining Enhanced Graph Attention Networks and Bidirectional Long Short-Term Memory to address this challenge. First, E-GAT captures structural dependencies between emotional features by constructing a semantic graph from text embeddings. Second, Bi-LSTM models temporal dynamics of sequential data, enabling effective integration of contextual information. We evaluated the model on three benchmark datasets: SemEval-2018 (text-only), RAVDESS (audio-visual), and CMU-MOSEI (multi-modal). Experimental results show that the proposed model achieves state-of-the-art performance: 58.5% accuracy and 68.7% F1-score on SemEval-2018, outperforming baseline models. On multi-modal datasets, it achieves 78.9% accuracy (RAVDESS) and 82.3% accuracy (CMU-MOSEI), demonstrating robust cross-modal generalization. This work advances emotion recognition by providing a unified framework for both text-only and multi-modal scenarios, with applications in human-computer interaction and mental health monitoring.
3 May 2026
Read appraisal →JMIR formative research
Comparing Perceptions of ChatGPT Use in Health Attitude Contexts Among Users and Nonusers: Cross-Sectional Study
BACKGROUND: In light of the growing use of artificial intelligence (AI) in health care, individuals' access to and use of health information are transforming. ChatGPT, an AI chatbot, provides immediate responses to health queries, with the potential to influence health-related attitudes, thereby raising concerns related to privacy, reliability, and security. OBJECTIVE: This study aimed to investigate the perceived usefulness, risks, anxiety, and social influence of ChatGPT on health attitudes among users and nonusers in Saudi Arabia. METHODS: A cross-sectional study was conducted using an online survey based on a validated tool. In total, 337 participants aged 18 years and older responded to questions assessing their perceptions of ChatGPT on health-related attitudes. RESULTS: Data showed that 76.1% (194/255) of the respondents used ChatGPT, with the majority being younger and more highly educated. The main uses for health-related purposes were health education (43/194, 22.2%) and physical activity guidance (31/194, 16%). The analysis showed that users considered ChatGPT useful for health-related decisions, with 45.9% (89/194) finding it easy to learn and use, but concerns about privacy (106/194, 54.7%) and reliability (87/194, 44.9%) remained. Among nonusers, security risks (39/61, 63.9%) were the major barrier to using AI-based tools for health purposes, and 68.9% (42/61) found such tools attractive and engaging. There were no statistically significant differences between users and nonusers across all examined sociodemographic characteristics (P>.05). CONCLUSIONS: The study established the potential of ChatGPT in improving health decision-making and revealed cultural, privacy, and trust issues that may affect its implementation. These findings underscore the importance of strengthening the security of AI-based applications to enhance public acceptability of related health policies and to support the safe integration of tools such as ChatGPT into the health care system.
3 May 2026
Read appraisal →JMIR public health and surveillance
Integrating AI Into Governmental Public Health Decision Making: Challenges, Considerations, and a Path Forward
Public health emergencies such as pandemics, natural disasters, and epidemics may require rapid, high-stakes decisions often made by elected officials with limited public health training. Artificial intelligence (AI) holds significant promise to enhance the quality, transparency, and timeliness of governmental decision-making during such crises. This paper examines the potential of AI as a decision-support tool for elected officials while identifying key technical, logistical, ethical, and policy challenges. Technical considerations include model accuracy, data representativeness, and privacy protection, while ethical imperatives center on fairness, transparency, and accountability to prevent amplification of existing health disparities. The paper further explores workforce development needs, emphasizing AI literacy and cross-sector collaboration to enable informed use of AI insights. This viewpoint presents a novel AI Decision Support Lifecycle framework specifically designed for governmental public health emergency response, mapping six phases from problem definition through post-emergency evaluation. We provide stakeholder-specific recommendations for model developers, health agencies, and elected officials, and illustrate practical application through a detailed case example and use cases. Drawing on empirical evidence regarding digital health technologies and AI governance, we emphasize that technology deployment alone is insufficient. Successful implementation requires complementary investments in organizational capacity, data infrastructure, workforce training, community engagement, and continuous evaluation. AI integration also requires robust governance frameworks, continuous model evaluation, and alignment with existing crisis management structures. Policy recommendations highlight the importance of ethical AI frameworks, risk assessments, and public engagement to foster trust. Ultimately, AI can strengthen public health decision-making if developed and implemented responsibly within transparent and equitable systems.
2 May 2026
Read appraisal →Child's nervous system : ChNS : official journal of the International Society for Pediatric Neurosurgery
Structural equation modeling of findings in pediatric tele-neuroradiology: a 2-year nationwide study of turnaround time predictors and clinical decision support
BACKGROUND: Pediatric neuroradiology faces significant workforce shortages, with teleradiology emerging as a vital solution. However, studies investigating findings and their operational impact from teleradiology centers remain limited. OBJECTIVES: To develop and validate structural equation models for identifying predictors of turnaround time and formulate recommendations for workflow optimization in pediatric tele-neuroradiology services. DESIGN: A retrospective cohort study following STROBE guidelines. SETTINGS: 107 hospitals across 17 states in the United States (US) via a teleradiology platform providing interpretation services through US board-certified radiologists. PATIENTS AND METHODS: We analyzed 9985 pediatric neuroradiology scans from 7958 patients (January 2023-December 2024). We utilized confirmatory factor analysis to validate findings structures, followed by structural equation modeling to predict turnaround times. Binary logistic regression models were developed with area-under-the-curve (AUC) estimation for performance assessment. Bootstrap validation with 5000 samples ensured model stability. MAIN OUTCOME MEASURES: Primary outcome was turnaround time. Secondary outcomes included requirements for multiple imaging studies, follow-up recommendations, and consultations. SAMPLE SIZE: A total of 9985 studies providing over 99% statistical power for detecting significant relationships. RESULTS: Factor analysis demonstrated a two-factor structure (trauma: α = 0.742, structural: α = 0.685). The structural model explained 7.8% of turnaround time variance, with computed tomography (CT) modality (β = -0.164), trauma score (β = 0.125), and structural score (β = 0.142) as significant predictors. Among immediate neurosurgical emergencies (n = 180, 1.8%), 89.4% achieved turnaround time within the 60-min benchmark for time-sensitive consultations. Prediction models demonstrated excellent discrimination: traumatic findings (AUC = 0.91), structural findings (AUC = 0.92), critical findings (AUC = 0.95), and a dedicated neurosurgical emergency model (AUC = 0.94, NPV = 0.996). A severity classification system showed strong validation against imaging needs (AUC = 0.76) and consultations (AUC = 0.89). CONCLUSIONS: Our study establishes a validated SEM framework for pediatric tele-neuroradiology with excellent predictive performance (AUC = 0.91-0.95). Among immediate neurosurgical emergencies (n = 180, 1.8%), 89.4% met the 60-min benchmark, and a dedicated emergency prediction model achieved AUC = 0.94. However, translation to improved neurosurgical care delivery and patient outcomes remains unvalidated, representing the next investigational priorities. LIMITATIONS: Retrospective design limits causal inference; a single platform may limit generalizability; the CT majority (96.8%) limits magnetic resonance imaging conclusions. Critically, post-diagnostic clinical outcomes, including neurosurgical consultations, interventions performed, and patient outcomes, were not tracked, precluding conclusions about whether documented operational efficiency translated to improved neurosurgical care delivery.
2 May 2026
Read appraisal →JMIR dermatology
Public Interest in Janus Kinase (JAK) Inhibitors for Alopecia Areata: A Google Trend Analysis
Public interest in Janus kinase (JAK) inhibitors for alopecia areata increased following media coverage and the 2022 US Food and Drug Administration (FDA) approval of baricitinib, highlighting the need for patient education and physician guidance on appropriate indications and treatment selection for hair loss disorders.
28 Apr 2026
Read appraisal →Biomedical physics & engineering express
Kali MC: an open-source toolkit for intraoperative electron radiation therapy treatment planning
Intraoperative electron radiation therapy (IOERT) is commonly delivered without intraoperative imaging, limiting patient-specific planning and requiring fast, reliable treatment calculations in the operating room. As a result, monitor unit (MU) calculations are often performed using spreadsheets and static look-up tables derived from measurements in water. We present Kali MC, an open source, Python-based software developed to streamline IOERT absorbed dose visualization, MU calculation and report generation for treatments delivered with the Liac HWL mobile accelerator. The software provides a graphical interface to visualize precomputed Monte Carlo absorbed dose distributions in water and performs MU calculations using experimentally determined output factors. Optional atmospheric pressure correction and predefined rescaling factors are supported to improve consistency of dose delivery and target coverage with the prescription isodose. Kali MC also generates customizable PDF treatment reports and exports DICOM RT Plan objects to facilitate integration with record-and-verify systems. By integrating validated dosimetric data, correction methods, and 3D dose visualization into a transparent platform, Kali MC supports efficient intraoperative decision-making; however, its use should be locally commissioned and evaluated according to institution-specific procedures and equipment characteristics.
28 Apr 2026
Read appraisal →BMJ health & care informatics
Does the accuracy of medication administration documentation improve with electronic medication systems? A stepped-wedge cluster randomised trial.
OBJECTIVES: Accurate documentation of medication administration is crucial for patient safety. This study compares the accuracy of documentation between paper-based and electronic medication management (eMM) systems. METHODS: A secondary analysis of data from a stepped-wedge cluster randomised trial of an eMM implementation was conducted. The eMM module allowed electronic prescribing, recording of drug dispensing, drug administration and medication reconciliation and monitoring. Medication administration processes were observed in a tertiary paediatric hospital in Sydney, Australia. Trained nurse observers recorded details of medication administrations which were then compared with the information documented by nurses about the administrations. RESULTS: 5095 administrations were analysed. With eMM compared with paper, significantly fewer administrations had no time documented (adjusted OR (aOR) 0.15; 95% CI 0.07 to 0.34; p<0.001), but documented details were more often inaccurate (aOR 4.6; 95% CI 3.3 to 6.6). Fewer doses were administered from incorrect prescriptions with eMM (aOR 0.25; 95% CI 0.18 to 0.36). CONCLUSION: eMM improves certain aspects of documentation but introduces new challenges, particularly in accurately recording administration details. Further investigation is needed to optimise these systems. TRIAL REGISTRATION NUMBER: Australian New Zealand Clinical Trials Registry 370325.
27 Apr 2026
Read appraisal →BMJ health & care informatics
AI-generated clinical summaries: errors and susceptibility to speech and speaker variability
OBJECTIVES: To evaluate whether variability in patients' communication style (personality), international English-accents (human and synthetic) and speech impairments affects the accuracy of a Clinical AI Scribe (CAIS) and identify where performance degrades to inform pre-deployment validation and monitoring. METHODS: We conducted simulated primary-care consultations using trained actors. For personality types, four scenarios were enacted, each with five patient-personality types. For accents, transcripts of consultations were used to generate combinations of seven accents across five scenarios. The CAIS produced summaries that were compared with transcripts, and errors classified as omissions, factual inaccuracies or hallucinations. For speech impairments, public recordings representing five profiles were transcribed and word-recognition accuracy calculated. RESULTS: Personality types showed no statistically significant differences in errors (all p>0.05). Extraversion had the highest total errors (median 3.5). Across accents, comparisons were non-significant for both patient and doctor voices (patients: p=0.851; doctors: p=0.980). Omissions predominated, with low rates of hallucinations and factual inaccuracies. Omissions were slightly higher for Chinese-accented and Indian-accented doctors (both medians 3.0). Conversely, speech impairments differed: cleft palate and vowel disorders were near-perfect, whereas phonological impairment markedly reduced recognition (p<0.001). DISCUSSION: Operationally, CAIS deployment should include clinician-in-the-loop verification, subgroup performance monitoring (accents, impairments) and predefined 'switch-off' criteria for severe phonological patterns. High-quality synthetic voices are a pragmatic proxy for accent testing when balanced corpora are unavailable. CONCLUSIONS: Under controlled conditions CAIS performance was broadly stable across communication styles and most accents, but vulnerable to specific speech characteristics, particularly phonological impairment, in this single-system simulation study.
27 Apr 2026
Read appraisal →American Society of Clinical Oncology educational book. American Society of Clinical Oncology. Annual Meeting
Tech That Scales: A Practical Framework for Artificial Intelligence-Enabled Cancer Care in Low- and Middle-Income Countries and Underserved US Counties
Cancer outcomes remain starkly unequal: 5-year survival rates for common malignancies in low- and middle-income countries (LMICs) often lag 20-40 percentage points behind high-income benchmarks, and similar disparities persist between well-resourced metropolitan centers and rural or safety-net settings within high-income countries. The gap is driven less by the absence of effective interventions than by workforce shortages, fragmented referral pathways, limited infrastructure, and loss to follow-up after abnormal screens. Technologies that scale must therefore function as deployable workflows integrating staffing, logistics, quality assurance (QA), governance, and monitoring not merely as stand-alone algorithms or devices.1 This review synthesizes evidence across four complementary technology families that address these constraints across the continuum of care: artificial intelligence (AI)-supported screening and triage as the population entry-point layer; Project ECHO telementoring as the workforce-capacity layer; electronic patient-reported outcomes and remote symptom monitoring as the longitudinal continuity layer; and AI-powered clinical trial prescreening hubs as the access-to-innovation layer. The technologies do not carry equal evidentiary weight, and they should not be deployed identically in every setting. Our aim is to show how oncology leaders can sequence them pragmatically inside a common operating logic while adapting to local infrastructure and governance. We organize the framework in patient journey order entry into care through screening, workforce support through telementoring, continuity through remote monitoring, and access to innovation through trial prescreening, and draw examples from both LMIC programs and underserved US settings. We therefore present a practical implementation playbook, including a 90-day launch checklist, staffing models, QA frameworks, equity and bias monitoring metrics, and a program outcomes dashboard designed for oncology leaders seeking to move beyond pilots toward durable, monitored deployment.
27 Apr 2026
Read appraisal →Journal of medical Internet research
Exploring Attitudes Toward AI-Based Contactless Sensors in Health Among Five Stakeholder Groups: Qualitative Study
BACKGROUND: The rapid rise of artificial intelligence-based contactless sensors (AI-CS) is expected to significantly transform how patients are measured, monitored, and understood through a versatile, noninvasive approach to data collection and health assessment. However, there is a lack of empirical research specifically focusing on AI-CS in health. Moreover, existing studies tend to focus on medical or patient perspectives, while neglecting other stakeholders such as researchers, political actors, or the general public. OBJECTIVE: The study aims to provide an in-depth empirical ethical analysis and, through a multistakeholder approach, a uniquely comprehensive overview by addressing the research question: what are the attitudes of different stakeholders (patients, health care professionals, researchers, political stakeholders, and the general public) toward AI-CS and their applications in health? METHODS: We conducted a cross-sectional study with 104 participants using a semistructured interview guide. Interviews were analyzed using qualitative content analysis with ATLAS.ti software (ATLAS.ti Scientific Software Development GmbH), following a 3-component model of feelings, thoughts, and behavioral aspects. RESULTS: The results of the study provide an in-depth analysis of attitudes toward AI-CS in health among different stakeholders. Overall, the results show a high level of openness to AI-CS in health across all stakeholder groups. In terms of feelings and their correlation with behavioral aspects, 2 key trends emerged: first, greater experience and knowledge correlated with a reduced tendency to react emotionally. Second, participants with positive experiences with technologies were generally more open and positive toward contactless sensors. The combined findings on thoughts and behavioral aspects highlighted 3 key tensions-around contact(lessness) and the importance and ambivalence of touch, between protection and surveillance (particularly regarding path- and context-dependency) and between the benefits and challenges of unobtrusiveness (especially in relation to control and governance implications). In addition, the analysis revealed the need for information and consent about AI-CS and clarified possible technical implementations and fields of application. CONCLUSIONS: This study provides a comprehensive and empirically grounded ethical analysis of stakeholder attitudes toward AI-CS in health. The findings offer valuable guidance for the responsible development, implementation, and governance of AI-CS in health care contexts.
27 Apr 2026
Read appraisal →BMJ health & care informatics
Using a large language model artificial intelligence agent to improve the efficiency of clinical quality measure evidence evaluation: a case study.
OBJECTIVES: To evaluate the feasibility and performance of a large language model (LLM)-based artificial intelligence (AI) agent, implemented within a structured Claim-Argument-Evidence System (CAES), for supporting the review of clinical quality measure (CQM) evidence in the Centers for Medicare & Medicaid Services Consensus-Based Entity (CBE) endorsement process. METHODS: The CBE conducted a pilot study using a previously endorsed measure. CAES extracted claims and citations from a submitted diagnostic performance measure for pneumonia, automatically retrieved additional relevant evidence from PubMed abstracts and assessed the quality, confidence and agreement of evidence supporting each claim. The system's assessments were compared with the judgement of a subject matter expert (SME). RESULTS: CAES completed the assessment in approximately 5 hours. The SME agreed with the CAES-assigned claim statuses for 69% of claims, was neutral for 11% and disagreed for 14%. Disagreements primarily stemmed from the need for contextual interpretation beyond abstracts. DISCUSSION: Manual evaluation of CQM evidence requires significant time and resources, estimated at over 2400 labour hours per review cycle, limiting efficiency and transparency.The AI agent evaluated 64 claims and 355 claim-evidence pairs related to the pneumonia diagnosis measure. It assigned claim statuses based on evidence strength and generated justifications. CONCLUSION: This pilot demonstrated the feasibility and potential of LLM-based AI agents to improve the efficiency and transparency of evidence review for CQMs. Further development is needed to incorporate additional data sources and extend applicability across the measure development lifecycle.
27 Apr 2026
Read appraisal →JMIR formative research
Exploring User Experiences of an Augmented Reality Smartphone App Prescribing Exercise for Children and Young People With Cancer: Results From a Qualitative Study
BACKGROUND: Mobile health (mHealth), and specifically smartphone apps, have grown exponentially in both functionality and accessibility and are becoming an important component of health care. Research exploring the use of mHealth for managing or treating chronic diseases, such as cancer, has shown promising effects. Yet, comparatively little work has examined how such technologies can enhance exercise interventions for young people with cancer. To optimize the effectiveness of mHealth in these contexts, it is essential to build a stronger evidence base on user experience. OBJECTIVE: This study aimed to investigate how healthy children and young people engaged with an augmented reality (AR) app developed specifically for children and young people undergoing cancer treatment, and to identify design features that may support engagement and behavior change in the intended clinical population. METHODS: School and university students, aged 8-21 years, were eligible to participate in the study. Practical workshops allowed participants to engage with the AR exercise app before taking part in focus groups to explore user experiences. Data were analyzed using qualitative content analysis, which also involved a critical friend approach using 2 researchers (HM and KS). Suggested improvements were mapped against the motivational affordances' taxonomy. RESULTS: A total of 39 participants aged 8-21 years took part in the focus group study. Participants found the demonstrations and varied exercises useful but expressed some concerns regarding data safety and functionality of the novel AR avatar. It was proposed that additional educational components, challenges, and rewards, as well as a customizable avatar, social support features, and audio instructions for a more inclusive design would be desirable and could enhance user experience. When mapped against the motivational affordances taxonomy, the suggested improvements aligned primarily with mechanisms of user education, challenges, feedback, cooperation, and comparison. CONCLUSIONS: This study provides an understanding of how apps that prescribe exercise can be optimized to enhance motivation and user experience. By assessing feedback and suggestions for improvements, the findings highlight key design features that may support engagement. While this initial work focused on healthy, age-matched participants, further evidence specifically in children and young people with a childhood cancer diagnosis is needed.
26 Apr 2026
Read appraisal →European journal of hospital pharmacy : science and practice
How to manage the overwhelming amount of database-detected interactions? A focus group meeting study on the management of database-detected risks of drug interactions.
OBJECTIVES: Database-assisted screening may help detect the risks of drug interactions, but less than 10% of flagged interactions have clinical manifestations, potentially leading to alert fatigue among healthcare providers. To address this issue, this study aimed to develop a concise framework to support the management of database-indicated interaction risks. METHODS: A grounded theory approach was used to design, evaluate and develop the framework. Participants included healthcare professionals and laypersons with experience using narrow therapeutic index medications, as well as individuals recommended by them. Two focus group discussions were conducted (N1=6 and N2=5), each lasting approximately 2 hours and following the same agenda. In one case, a follow-up one-on-one interview was held because the participant indicated further insights at the end of the session. RESULTS: Participants identified key decision-making questions for the framework, including the evidence base for interacting drugs, severity and documentation level of the interaction, availability of safer alternatives, and the presence of confounding factors. The framework was considered useful and, following refinement, potentially suitable for clinical implementation. A key insight was that integration into healthcare curricula is essential for achieving long-term impact. Overall, the proposed tool may assist in managing interaction risks in diverse clinical scenarios and reduce alert fatigue among healthcare professionals. CONCLUSION: Although further clinical validation is needed, the framework provides a foundation for improving the management of database-detected drug interactions.
26 Apr 2026
Read appraisal →Journal of medical Internet research
Perspectives Regarding the Privacy, Security, and Confidentiality of Data Collected via mHealth Apps in Saudi Arabia: Qualitative Analysis
BACKGROUND: As Saudi Arabia aims to enhance the adoption and utilization of mobile health (mHealth) apps in the health care sector, it is vital to identify the potential challenges that may be faced in the process and how to address them. OBJECTIVE: This study aimed to explore patients' and stakeholders' perspectives regarding the privacy, security, and confidentiality of data collected via mHealth apps in Saudi Arabia. METHODS: A qualitative research design was used by conducting an in-depth interview with 25 participants, comprising health care workers (HCWs), patients, and mHealth app developers. The interview questions focused on awareness of mHealth apps and issues relating to data security, privacy and confidentiality, benefits of mHealth apps, challenges faced, and features that may improve the utilization of such digital health technologies. Thematic analysis was performed using NVivo software (version 12; QSR International). RESULTS: Patients were relatively less informed regarding events associated with data privacy and security than HCWs. Specific factors were identified to influence patient awareness with HCWs, suggesting the need for patient education and collaboration with cybersecurity organizations. Participants posited that advanced security features, user-friendly features, online consultation for emergencies, remote monitoring features, and taking patient needs into account are important facilitators of mHealth apps. Shared experiences mainly revolved around the accessibility to information and reliability of mHealth apps, while the challenges were related to poor usability, technical difficulties, data security, and data breaches. Patient autonomy, remote monitoring, self-care management, medical adherence, and time and cost-saving summed up the perceived benefits of the mHealth app. CONCLUSIONS: These findings may assist policymakers in developing strategies to improve Saudi users' and patients' adoption of mHealth apps and address the concerns raised to benefit significantly from these advanced health care modalities.
25 Apr 2026
Read appraisal →PloS one
Cognitive load and pedagogical tension in multi-platform online learning: Evidence from Chinese higher education
The proliferation of digital tools has transformed higher education into a complex, multi-platform online learning environment. This study investigates the paradoxical effects of this multi-platform environment on the student experience within Chinese higher education. Through the qualitative and quantitative data from a cross-sectional survey of 8,616 university students, we analyzed platform usage patterns, perceived benefits and challenges. The findings reveal a significant paradox: while students express positive attitudes towards platforms' educational value, acknowledging benefits like enhanced self-directed learning and improved teacher-student interaction, they concurrently report substantial practical burdens. A serial mediation model confirmed that platform multiplicity increases extraneous cognitive load, which in turn elevates tool fatigue, ultimately leading to a more negative perception of the learning experience. This study identifies a core "pedagogical tension" between the intended benefits of educational technology and the lived reality of a fragmented, high-friction user experience that encourages instrumental engagement over deep learning. These findings underscore the urgent need for institutions and instructors to adopt a more strategic, integrated approach to educational technology to reduce fragmentation and prioritize a seamless student learning experience.
25 Apr 2026
Read appraisal →BMJ health & care informatics
Barriers associated with the implementation, adoption, scale-up and sustainability of mHealth in Sub-Saharan Africa: a systematic review guided by the NASSS Framework
OBJECTIVES: The objectives of this study were to highlight and evaluate the factors affecting the non-adoption, abandonment of and challenges to scale-up, spread and sustainability of mHealth in Sub-Saharan Africa (SSA) and to assess the appropriateness of the Non-adoption, Abandonment, Scale-up, Spread, and Sustainability (NASSS) Framework in the SSA context. METHODS: A systematic search of published research was conducted using PubMed and Embase research databases from 2012 to 2025. Peer-reviewed studies from SSA countries, evaluating mHealth interventions, were included in this study. Papers were analysed to identify barriers according to the seven domains of the NASSS Framework, which are (1) condition, (2) technology, (3) value proposition, (4) adopters, (5) organisation(s), (6) wider system, (7) embedding and adaptation over time RESULTS: This study identified 409 barriers across 91 papers. 309 barriers were covered by the seven NASSS Framework domains, while 100 were unaddressed by the framework. The main barriers were associated with domain 2, the technical aspects of the technology (27%). Other prominent barriers related to the wider system (18%), the adopter (11%), the organisation (9%) or condition (8%), with the value proposition (1%) and embedding and adaptation over time (1%), least hindering implementation. Challenges which were unaddressed by the framework (24%) included socioeconomic and infrastructural issues, distinct to lower and middle-income countries (LMIC). DISCUSSION: This paper highlights the importance of evaluating rudimentary aspects of mHealth and preparing infrastructure before addressing more complex issues during mHealth implementation in SSA. CONCLUSION: Although the NASSS Framework is applicable to the SSA region, it would benefit from some further development to improve its applicability in LMICs.
25 Apr 2026
Read appraisal →JMIR human factors
Psychometric Properties of Digital Divide Measurement Instruments in Health Contexts: Systematic Review
BACKGROUND: The digital divide constitutes a growing form of social inequality, influencing access to, use of, and benefit from digital health technologies. Accurate assessment of this phenomenon requires psychometrically sound measurement instruments, especially in health care contexts. OBJECTIVE: This systematic review aimed to identify and evaluate the psychometric properties of instruments developed to measure the digital divide, with a particular focus on their application in health-related settings. METHODS: A systematic review was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 and COSMIN (Consensus-Based Standards for the Selection of Health Measurement Instruments) guidelines. Literature searches were performed in PubMed, Scopus, Dialnet, and Web of Science for studies published between 2015 and 2025. Inclusion criteria required studies to report a psychometric evaluation of instruments measuring the digital divide. RESULTS: Four studies met the inclusion criteria (ranging from 875 to 1337 participants). The identified scales primarily measured digital skills and capital. All studies reported adequate internal consistency, with Cronbach α coefficients ranging from 0.70 to 0.94. Structural validity was confirmed through factor analysis in all studies. However, data on temporal stability and responsiveness were absent. CONCLUSIONS: Existing instruments show potential for assessing digital inequality in health, but further psychometric validation and cultural adaptation are required. Nursing professionals are encouraged to lead efforts in developing and validating context-sensitive instruments to promote digital health equity.
25 Apr 2026
Read appraisal →BMJ open
Course of psychological symptoms and the initial management strategies in general practice using electronic health records in the Netherlands
BACKGROUND: Patients who consult their GP about psychological complaints, such as feeling anxious or depressed, are often initially given a psychological symptom diagnosis. However, it remains unclear whether these symptoms will develop into psychiatric conditions, which is crucial for informing patients about their prognosis and guiding GPs in their management. AIMS: To explore the course of psychological symptom diagnoses and compare GPs' management strategies for (a) psychological symptom diagnoses that persisted for more than a year and (b) those that changed into psychiatric conditions during the first year of care. METHODS: We performed a retrospective cohort study using the Family-Medicine Network database. We included all episodes of care (EoC) that started with a psychological symptom diagnosis between 2008 and 2021. We performed negative binomial analyses and logistic regression analyses to compare management strategies and number of contacts between EoC that changed into psychiatric conditions and persistent psychological symptoms (>1 year) during the first year of care. RESULTS: Out of the 14 633 EoC that started with a psychological symptom diagnosis, 79.4% resolved within 1 year, 12.8% persisted as psychological symptoms and 7.8% changed into psychiatric conditions. In EoC that changed into psychiatric conditions, as compared with EoC that persisted as psychological symptoms, we observed a significantly higher number of contacts with the GP (RR=1.76, 95% CI 1.63 to 1.91) as well as an increased total number of interventions (RR=1.71, 95% CI 1.58 to 1.84). CONCLUSION: Most psychological symptoms remain for only a short period of time and only a few persist or change into psychiatric conditions. Future research should investigate the factors that influence patients' decisions to seek further help from their GPs as well as those that contribute to the transition from psychological symptom diagnoses to a psychiatric condition.
25 Apr 2026
Read appraisal →Nicotine & tobacco research : official journal of the Society for Research on Nicotine and Tobacco
Computer Vision Models for Detecting Large Cigars on Social Media
INTRODUCTION: Large cigars, which include premium cigars and large manufactured cigars, are widely promoted on social media and digital media, highlighting the critical need for scalable methods to monitor such content. To fill this need, we aimed to train and test an automated computer vision model to identify large cigars. METHODS: In 2022, we obtained 876 large cigar images and 1526 non-cigar images (ie, cigarettes, e-cigarettes, and pens) from Reddit, a global social media platform organized into subcommunities (subreddits) focused on specific topic areas, discussions, and communities. We used the You Only Look Once Version 7 (YOLOv7) computer vision model to train large cigar detection and evaluated its performance using recall, precision, and F1 scores (ie, harmonic mean of recall and precision scores). The data were split into 75% (594 cigar images and 250 non-cigar images) for training and 25% (282 cigar images) for validating and testing. Non-cigar images (n = 1276) were used to evaluate for false positives. RESULTS: Our model achieved a recall of 0.98, a precision of 0.99, and an F1 score of 0.98, demonstrating high accuracy in correctly identifying large cigars while minimizing false positives. It successfully distinguished large cigars from similarly shaped objects with an accuracy of 97%. CONCLUSIONS: Our computer vision model accurately distinguished large cigars from non-cigar images, demonstrating its potential as an automated, scalable tool for monitoring this understudied tobacco product on social media and other digital media. IMPLICATIONS: This computer vision model offers a novel and scalable approach to tobacco monitoring by automatically detecting large cigars in social media content. Its application may enhance the efficiency of product identification and facilitate research on use patterns, user perceptions, and emerging marketing strategies, thereby generating evidence to inform regulatory decision-making and policies aimed at preventing initiation.
24 Apr 2026
Read appraisal →Journal of the American College of Cardiology
Using Artificial Intelligence to Adjudicate Major Adverse Cardiovascular Events in Clinical Trials
BACKGROUND: Major adverse cardiovascular events (MACE)-cardiovascular (CV) death, nonfatal myocardial infarction (MI), and nonfatal stroke-are highly relevant clinical outcomes. In global randomized trials, medical records review by a physician clinical events committee (CEC) is the conventional standard for adjudicating MACE but is labor intensive. Automated adjudication with the use of artificial intelligence (AI) could reduce cost and improve reproducibility. OBJECTIVES: In this study, the authors sought to develop and validate an AI-based adjudication system for MACE and compare its performance with CEC adjudication in a large global randomized trial. METHODS: We developed an AI-based system ("Auto-MACE") that uses an iteratively refined prompt of the OpenAI o1-mini language model to adjudicate MACE events, and a Clinical Longformer model trained on adjudicated events to assign a confidence level. We validated Auto-MACE against CEC adjudication in the PARADISE-MI global clinical trial comparing sacubitril/valsartan and ramipril in 5,661 patients with MI complicated by systolic dysfunction or pulmonary congestion. RESULTS: Auto-MACE provided a confident adjudication in 315/455 deaths (69%), 301/659 potential MIs (46%), and 136/167 potential strokes (81%). Auto-MACE agreed with the CEC adjudication in 97%, 89%, and 88% of confident events, respectively. Among all events, Auto-MACE agreed with CEC adjudications in 86% of deaths, 76% of potential MIs, and 84% of potential strokes. The estimated effect of sacubitril/valsartan vs ramipril on composite MACE was similar with Auto-MACE adjudication (HR: 0.91; 95% CI: 0.78-1.07) and CEC adjudication (HR: 0.90; 95% CI: 0.77-1.05). CONCLUSIONS: AI-based adjudication of MACE showed high agreement with human CEC adjudication, especially for CV death and stroke, and where the model was confident. Initial AI-based adjudication with CEC review of uncertain events may reduce workload while maintaining accuracy.
24 Apr 2026
Read appraisal →Journal of medical Internet research
Exploring Patient Perspectives on the Use of Artificial Intelligence to Inform Joint Decision-Making for Patients With Multiple Conditions in Primary Care in the United Kingdom: Qualitative Study
BACKGROUND: Multimorbidity, living with 2 or more long-term health conditions, is increasing globally and now affects over one-quarter of adults in England. People with multiple long-term conditions (MLTC) face complex health and treatment challenges, often experiencing fragmented care within systems oriented toward single-disease management. Artificial intelligence (AI) has the potential to support clinicians and patients by analyzing complex health data, optimizing treatment strategies, and predicting disease trajectories. OBJECTIVE: The OPTIMAL (Optimizing Therapies, Disease Trajectories, and AI-Assisted Clinical Management for Patients Living with Complex Multimorbidity) project aims to develop AI-enabled tools to support shared decision-making in primary care. This study explored how patients with MLTC perceive the use of AI to inform joint decision-making in primary care. METHODS: Semistructured interviews were conducted via telephone or video call with 29 adults living with MLTC between July and November 2023. Participants were recruited through general practitioner practices via the Clinical Practice Research Datalink and community-based organizations across the West Midlands. Interviews were transcribed verbatim and analyzed thematically using an inductive approach. Members of a patient advisory group were involved in developing study materials, refining the interview guide, and reviewing emerging findings to ensure relevance and authenticity. RESULTS: Participants identified potential benefits of AI in enhancing consultation efficiency and accuracy, improving access to information for patients and clinicians, promoting early detection of health changes, and reducing health care inequalities. However, concerns were raised about the loss of human interaction, data privacy and security, transparency of algorithms, and the potential for bias and inequity in AI systems. Trust and acceptance varied by age and familiarity with technology. Some participants expressed uncertainty about what AI entails and how it could be used in primary care. CONCLUSIONS: Patients with MLTC viewed AI-assisted decision-making in primary care with cautious optimism. While many recognized potential benefits for coordination and personalization of care, others expressed reservations about privacy, fairness, and the risk of diminished human connection.
24 Apr 2026
Read appraisal →JMIR formative research
Multimodal Sentiment and Emotion Analysis Framework for Personalized Health Coaching Messages: Proof-of-Concept Study
BACKGROUND: Text generation approaches in health care communication have evolved along 2 major paths. The first path involves generative adversarial networks, progressing from basic architectures to specialized variants like Text-to-Text Generative Adversarial Network (TT-GAN) and Time and Frequency Domain-Based Generative Adversarial Network (TF-GAN), which address challenges in discrete text generation through techniques such as Gumbel-Softmax and reinforcement learning. The second path emerges from transformer-based architectures, particularly Generative Pretrained Transformer-2 (GPT-2), which uses extensive pretraining and self-attention mechanisms to generate contextually appropriate text. GPT-2's transformer architecture enhances persuasive health communication by generating personalized messages using various strategies like task support, dialogue support, and social support for effective health interventions. OBJECTIVE: This study aimed to use GPT-2 as a generative method to construct persuasive text in a dataset and compare the performance of sentiment analysis and emotion detection analysis. METHODS: We combined sentiment analysis tools (VADER [Valence Aware Dictionary and Sentiment Reasoner] and TextBlob) with emotion detection methods (Text2Emotion and NRCLex [National Research Council Lexicon]) to analyze health coaching messages across different persuasive types: reminder, reward, suggestion, and praise. RESULTS: TextBlob and VADER achieved accuracies of 57% and 69%, respectively, while RoBERTa (robustly optimized BERT approach)-sentiment outperformed them with an accuracy of 88%. Emotion detection showed a high prevalence of "joy" and "happy" labels (93.69% positive skew). While transformers excel in accuracy, lexicon-based models like VADER offer a better performance-efficiency balance for real-time health communication systems. For emotion detection, all categories showed perfect accuracy (1.0), while trust showed mixed results, with precision, recall, and F1-score values ranging from 0.81 to 0.96. The emotion detection analysis revealed varying success rates across different emotions, with some categories, such as anger and neutral, showing reasonable performance and others, such as trust, showing mixed performance. CONCLUSIONS: This research contributes to understanding the emotional dynamics of persuasive health communication and highlights both the capabilities and limitations of current natural language processing tools in analyzing health-related persuasive messaging. This proof-of-concept study using synthetically generated data establishes a methodological framework for multimodal sentiment and emotion analysis. The findings require validation with real-world health coaching messages before clinical deployment.
24 Apr 2026
Read appraisal →JMIR medical informatics
Modeling Diabetes Risk and Progression With Public Health Data: Ontology-Guided, Simulation-Capable Digital Twin Study
BACKGROUND: Digital twins (DTs) offer a paradigm for health care by enabling data-driven, simulation-capable representations of individual health trajectories. However, DT development remains limited by the scarcity of standardized, temporally structured, and multidomain data suitable for modeling chronic disease progression. Most existing DT studies rely on narrowly scoped or proprietary datasets, restricting generalizability. Public health datasets, such as the Midlife in the United States study, provide rich biopsychosocial information but are underused due to structural complexity and lack of semantic integration frameworks. OBJECTIVE: This study aimed to develop and evaluate an ontology-guided, agent-orchestrated framework for constructing offline, simulation-capable, and progression-aware DTs from public health datasets. Using diabetes as a case study, the framework integrates agent-based orchestration, medical ontologies, and large language model (LLM)-assisted semantic reasoning with machine learning to support explainable feature structuring, risk prediction, and predictive "what-if" progression analysis. METHODS: A 6-stage DT framework was developed and applied to Midlife in the United States wave 2 (baseline) and wave 3 (follow-up) data. Ontology- and LLM-assisted feature selection identified predictors across biological, behavioral, psychosocial, and socioeconomic domains. Cleaned and harmonized data were used to train predictive models (random forest, eXtreme gradient boosting, and logistic regression) to estimate diabetes onset at follow-up. A state-transition simulator was implemented to model between-wave progression dynamics, quantify transitions across low-, medium-, and high-risk states, and evaluate predictive "what-if" scenarios such as weight reduction and lifestyle improvement. Model performance was assessed using accuracy, F1 score, area under the receiver operating characteristic curve (AUC), and calibration metrics. RESULTS: From 9976 candidate variables, ontology- and LLM-guided selection retained the top 200 relevant predictors spanning biological, behavioral, psychosocial, and socioeconomic domains. Predictive modeling achieved strong discrimination, with random forest (AUC=0.82, accuracy=0.76) and eXtreme gradient boosting (AUC=0.81, accuracy=0.75) outperforming logistic regression (AUC=0.78). The state-transition simulator reproduced realistic progression patterns: 33.9% (1414/4174) of participants changed risk states between waves, and the high-risk group increased from 10.8% (451/4174) to 32.2% (1344/4174). Next-state prediction accuracy reached 92.5%. Predictive "what-if" analyses showed that with a simulated 10% weight reduction, model-estimated diabetes cases decreased by 98 (from 576 to 478). A placebo test (0% weight change) produced less than 0.3% difference in risk distribution, confirming model stability. CONCLUSIONS: This study presents a foundational, ontology-guided, and agent-orchestrated framework for constructing offline, simulation-capable, and progression-aware DTs from public datasets. By combining semantic reasoning, multidomain predictors, and predictive "what-if" progression simulation, the framework transforms static population data into longitudinal, interpretable representations of individual health trajectories. The proof-of-concept application to diabetes demonstrates that public health data can support robust and explainable DT models for exploratory risk analysis and hypothesis generation, without implying causal intervention effects or direct clinical decision support.
24 Apr 2026
Read appraisal →Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer
Implementing an electronic health record workflow to identify parents with cancer and implications for supportive oncology care
PURPOSE: Parents with cancer face unique psychosocial challenges as they navigate treatment while caring for dependent children, yet parenthood status is rarely captured in oncology workflows. This initiative evaluated the feasibility and outcomes of a new electronic health record (EHR)-based workflow to systematically identify parents with cancer and then connect them to family-focused assessment and supportive care. METHODS: A two-phase quality improvement initiative was conducted at an NCI-designated comprehensive cancer center. Phase 1 included process mapping, analysis of tumor registry and distress screening data, and development of an EHR screening item assessing parenting status. Phase 2 implemented the item ("Patient is a parent/caregiver with child/ren < 22 years of age?") into oncology workflows, linking identified parents to a new supportive care program, Family Circle, which includes psychosocial assessment, navigation, and family-centered support. Training for oncology clinicians included in-service education. Analyses examined prevalence estimates and associations between parenting, distress, anxiety, and depression and reports of quality improvement metrics. RESULTS: Among N = 2,172 screened patients, 20.6% identified as parents; the age-adjusted prevalence of parenting with children under 21 was 16.9%, consistent with national estimates. Parents reported significantly higher distress (odds ratio [OR] = 1.35, 95% confidence interval [CI] = 1.11-1.64, p = 0.003) and anxiety (OR = 1.28, 95% CI = 1.06-1.52, p = 0.009) than matched non-parents. Parents with younger children exhibited greater anxiety than those with older children (F = 4.32, df = 2, 393, p = .01). During implementation, N = 431 parents were identified in 6 months (79.4% through the EHR item). CONCLUSION: Embedding parenting status screening into the EHR is feasible, accurate, and critical for linking parents with cancer to supportive care. Wider implementation of such workflows may enhance psychosocial oncology practice and improve cancer parenting outcomes.
24 Apr 2026
Read appraisal →American journal of health-system pharmacy : AJHP : official journal of the American Society of Health-System Pharmacists
Evaluating medication record creation before and after an electronic health record transition
PURPOSE: In 2022, St. Jude Children's Research Hospital underwent an electronic health record (EHR) transition, migrating from Cerner Millennium to Epic. This transition required a comprehensive revision of the medication build and configuration process and, subsequently, prompted an analysis of lessons learned during the transition. Through enhancement of the medication build and configuration process, the standardized procedure increases collaboration with subject matter experts and other departments within the institution as medication records are created. SUMMARY: The transition involved stakeholder meetings with members from operational leadership, informatics, medication safety, and pharmacy inventory management to update the existing process to support the new system. Through the creation of an initial version and subsequent iterations, the process was refined to efficiently support and control the quality of medication build requests. The updated process also maintained existing oversight over the entire process. The transition highlighted differences between the original EHR and the updated EHR in the management of medication builds. The differences between the EHRs and medication build processes presented a learning curve for individuals involved in medication builds and resulted in an initial lengthening of the turnaround time. Following iterative improvements to the usability of the medication build process and enhancements in automatic communication, the turnaround time substantially decreased. CONCLUSION: The insights gleaned from the transition process underscore the importance of aligning operational processes with user needs, fostering iterative improvements, prioritizing effective communication, and maintaining a delicate balance between urgency and accuracy in managing medication record creation and configuration within EHRs. This knowledge can be used to guide the creation of a medication build process or transition from an existing process.
24 Apr 2026
Read appraisal →Behavior research methods
A modified expectation-maximization algorithm for accelerated item response theory model estimation with large datasets
The expectation-maximization (EM) algorithm is widely used for parameter estimation in item response theory (IRT) modeling. However, when applied to datasets with large numbers of individuals and items, the standard EM algorithm can be slow to converge, with computationally expensive E-steps. We propose a modified EM algorithm to accelerate estimation for unidimensional two-parameter logistic IRT models. The modified algorithm uses a two-stage structure with partial-step updating over data subsets to reduce convergence time, while maintaining comparable accuracy and precision. The first two simulation studies evaluated its performance relative to standard EM, focusing on convergence time, parameter recovery, and standard error estimation across varying subset sizes and item counts. The third study demonstrated its scalability and runtime advantage in a large-scale testing scenario involving one million respondents and 100 items. The fourth study evaluated robustness under departures from unidimensionality. The proposed algorithm showed time reductions under smaller subsets with 36 items, consistent reductions across all subset sizes with 54 items, and the largest reduction (60%) with 40-item forms constructed from 100 items, while maintaining comparable estimation performance. These results highlight the algorithm's potential for large-scale applications involving tens of thousands of respondents and moderate-to-large item pools, with modifications that can be integrated into existing EM routines. We conclude with directions for operational use and potential extensions to multidimensional IRT estimation.
23 Apr 2026
Read appraisal →American journal of health-system pharmacy : AJHP : official journal of the American Society of Health-System Pharmacists
Implementation of dispense waste functionality with new electronic health record system.
PURPOSE: Practice changes were implemented and adjusted to ensure regulatory compliance with JZ modifier documentation. SUMMARY: The JW modifier allows institutions to bill for partially unused and discarded single-dose containers of drugs and biologics, whereas the JZ modifier is used to report that no amount of drug was discarded from a single-dose container. These modifiers are essential for accurate coding and resource stewardship, especially in pediatric hospitals, where medication doses are often patient specific and require manipulation of single-dose containers. Starting July 1, 2023, the Centers for Medicare & Medicaid Services required providers to report the JZ modifier on all claims that bill for drugs from single-dose containers. Our institution underwent a soft go-live in June 2023, which allowed the institution to make necessary process changes to ensure regulatory compliance. In June 2023, only 31.1% of waste was documented at the time of medication preparation. To capture waste documentation post preparation, a pharmacy technician was trained to review missed opportunities and document waste retrospectively. Monthly monitoring of waste documentation, including missed opportunities during drug preparation, was conducted and further staff education was provided. A reassessment of waste documentation was completed in February 2024, which highlighted that real-time waste documentation completed during compounding had increased to 85.9%. By documenting waste during drug preparation, the resources spent reviewing reports for drug preparations from the previous day and the need to document waste are reduced. CONCLUSION: Through process implementation and staff education, compliance with real-time waste documentation during drug preparation increased from 31.1% to 85.9% (P < 0.001) in 8 months.
23 Apr 2026
Read appraisal →Infectious diseases of poverty
Progress and challenges in development of minimum essential dataset for disease surveillance through a One Health lens: a scoping review
BACKGROUND: Minimum essential dataset (MED) enables One Health surveillance by facilitating cross-sectoral data sharing. Yet, existing MED research represents insufficient integration of the One Health concept, and its progress in disease surveillance remains unclear. This study aims to evaluate the current progress and gaps in MED research in disease surveillance through a One Health lens. METHODS: This scoping review systematically searched Embase, PubMed, Scopus, and Web of Science from the inception of databases to November 30th, 2024 to identify studies published in English. Studies were independently screened by two reviewers for inclusion based on their relevance to the development of MEDs for disease surveillance. The progress and challenges were synthesized based on data extracted from eligible studies. RESULTS: This review includes 28 eligible studies for analysis, all of which focused on infectious diseases. MED development is predominantly concentrated on the human health interface (89.3%), whereas animal and environmental interfaces are comparatively limited (≤ 10.7%). The One Health concept has been adopted in 16 of the eligible studies; however, there is still insufficient interdisciplinary and cross-sectoral collaboration (57.1%). Meanwhile, numerous methods have been employed in the development of MEDs, with qualitative approaches being the most prevalent (42.9%), while data-driven approaches remain scarce. The absence of a standardized approach is recognized as a primary barrier. Thus, we propose a pathway and operational tool to detail the specific steps, which could facilitate future research on MED. CONCLUSION: This review highlights critical gaps in MED development at the human-animal-environment interface, and proposes a pathway and operational tool to inform future development of MED, including scope identification, establishment, modification and improvement. This study promotes the improvement of disease surveillance system and provides insights to enhance future preparedness for reducing the global public health burden.
23 Apr 2026
Read appraisal →JMIR infodemiology
Cross-Sectional Evaluation of Medical Disinformation Safeguards in Consumer-Facing Large Language Model Platforms
This cross-sectional evaluation of six consumer-facing large language model platforms found significant heterogeneity in safeguard performance against the generation of health disinformation, with Claude and ChatGPT demonstrating complete resistance across all prompt types, while Copilot, Meta AI, Grok, and Gemini exhibited substantial vulnerabilities.
22 Apr 2026
Read appraisal →JMIR human factors
Perceptions of Health Care Professionals on the Integration and Use of AI in Clinical Cancer Care: Interview Study
BACKGROUND: Artificial intelligence (AI) is increasingly recognized for its potential to transform cancer care. However, much of the existing evidence of its efficacy comes from controlled settings. There remains a need to complement this knowledge with insights into how AI tools are perceived and used in real-world clinical settings, as well as how their use impacts clinical practice. OBJECTIVE: This study aimed to explore key factors influencing clinicians' acceptance of AI tools and examine how AI adoption and use impact clinical workflows in cancer care. METHODS: We used purposive sampling for recruiting oncology-related health care professionals and collected data using web-based semistructured interviews to gather their perceptions. Data were thematically analyzed and interpreted through the lenses of sociotechnical systems theory and the Unified Theory of Acceptance and Use of Technology. RESULTS: Participants largely accept and perceive AI tools as beneficial to clinical practice. Unified Theory of Acceptance and Use of Technology constructs were reflected in our data as determinants of intention to adopt AI tools. Trust appears as an influential factor in shaping attitudes toward AI tools. Acceptance is found to both precede AI tool use and to grow following successful integration. The use of AI tools is perceived to yield operational benefits, such as reduced workload and time savings, and clinical benefits, such as increased diagnostic reliability and reduced patient recall. Minimal disruption to clinical workflows following integration of AI tools was reported for some cancer screening applications and organ-at-risk segmentation, whereas greater disruption was anticipated for 3D cancer screening. Although accountability and lack of explainability are highlighted in literature as barriers to AI adoption, participants do not view these as significant obstacles in image-based diagnostic contexts. Additionally, negative impacts, such as overreliance on AI and reduced critical review of AI results, arise in association with the use of AI tools. CONCLUSIONS: Participants perceive AI tools to deliver benefits to clinical cancer care. However, their adoption relies on their alignment with clinical needs and seamless integration into clinical workflows. To encourage clinician acceptance, the identified concerns must be addressed. Future work should focus on training programs, co-design with clinicians, and exploration of mitigation strategies for emerging adverse effects, such as automation bias and potential skill erosion.
22 Apr 2026
Read appraisal →Scientific reports
End-to-end pipeline for automated heart failure diagnosis with clinical notes using SNOMED-CT
Diagnosis of heart failure is complex but crucial for patient outcomes and often hindered by the untapped potential of unstructured clinical notes. We introduce a novel end-to-end pipeline for heart failure diagnosis, leveraging electronic health records (EHR) and German clinical notes from 846 patients. Our pipeline synthesizes abbreviation disambiguation, translation of German clinical notes to English, medical entity linking to SNOMED-CT, and subsequent classification. The classification was performed using a Support Vector Machine (SVM) and compared against a fine-tuned medBERT.de neural baseline. We reduced the reliance on training data with zero-shot learning to address limitations with abbreviation disambiguation and entity linking approaches. Validation against benchmark datasets and cardiologists demonstrates high accuracy for real clinical use. Abbreviation disambiguation achieved an accuracy of up to 96.1%. Entity linking achieved competitive performance compared to state-of-the-art approaches on selected evaluation datasets. The SVM classification approach utilizing SNOMED-CT concepts and EHR data achieved an F1-score of 65.3%, on par with the medBERT.de neural baseline using clinical notes and EHR data. Despite challenges regarding limited language-specific resources and reference dataset availability for SNOMED-CT annotations in German, our pipeline demonstrates high potential for real-world clinical use and clinical decision support grounded in the standardized SNOMED-CT ontology.
21 Apr 2026
Read appraisal →PloS one
An explainable multi-head attention network for healthcare IoT threat detection based on the MedDefender-MHAN framework
The rapid proliferation of Internet of Medical Things (IoMT) devices in healthcare environments has created critical cybersecurity vulnerabilities that demand both accurate and interpretable intrusion detection solutions. Existing deep learning-based intrusion detection systems (IDS) achieve high detection accuracy but lack inherent explainability, limiting their clinical adoption under regulatory frameworks such as GDPR and FDA guidelines. This paper presents MedDefender-MHAN, an explainable multi-head attention network specifically designed for healthcare IoT threat detection. The proposed framework introduces a novel dual-stream architecture that combines convolutional neural networks for local spatial feature extraction with transformer-based encoders for long-range temporal dependency modeling. Unlike existing approaches that apply explainability as a post-hoc process, MedDefender-MHAN embeds interpretability directly into the multi-head attention mechanism, enabling real-time gradient-weighted explanation generation without external XAI pipelines. Evaluated on CICIDS2017 and TON_IoT benchmark datasets, MedDefender-MHAN achieves detection accuracies of 99.47% and 98.92% respectively, with sub-3ms inference latency and a throughput of 435 samples per second. Explainability evaluation demonstrates 94.6% alignment with expert-annotated attack signatures and 91.9% temporal accuracy, outperforming post-hoc methods such as SHAP and Integrated Gradients. These results confirm that MedDefender-MHAN provides a clinically viable, regulatory-compliant security solution for real-world healthcare IoT infrastructure. The proposed framework addresses the dual imperatives of methodological transparency and clinical impact, directly responding to the growing need for trustworthy AI-driven security solutions in regulated healthcare IoMT environments.
20 Apr 2026
Read appraisal →Journal of medical Internet research
The Real-Time Support Role of Augmented Reality Technology in Shared Decision-Making in Neurosurgery Under the SEGUE Framework: Randomized Controlled Trial
BACKGROUND: Preoperative risk communication is essential for shared decision-making (SDM) in neurosurgery; however, conveying complex neuroanatomy and surgical risks using traditional verbal explanations can limit understanding and contribute to dissatisfaction and medicolegal disputes. Augmented reality (AR) may provide patient-specific, interactive 3D visualization to support these conversations. OBJECTIVE: This study evaluates whether AR-assisted preoperative risk communication improves objective understanding and other SDM-related outcomes, compared with communication supported by a conventional physical anatomical model within a standardized SEGUE-informed protocol. METHODS: A prospective, single-center, randomized controlled trial was conducted with 62 neurosurgery communication recipients (patients when capable; otherwise, a legally authorized representative [LAR]). Patients were stratified by planned surgical approach (frontal, parietal, and occipital) and, within each stratum, were randomized into an experimental AR group and a control physical-model group. The primary outcome was postsession objective understanding, assessed by a multiple-choice knowledge questionnaire. Secondary outcomes were subjective understanding, communication satisfaction, pre-to-post anxiety changes, communication duration, and neurosurgeons' communication skills from video recordings. RESULTS: Of the 67 individuals screened, 62 communication recipients were enrolled and completed all assessments (patients, n=30; LARs, n=32; AR, n=32; and control, n=30). Objective understanding was higher with AR than with the physical model in the prespecified pooled comparison (P=.01). Communication satisfaction was also higher with AR (P<.001). There were no clear between-group differences in subjective understanding (P=.41), anxiety changes (ΔState-Trait Anxiety Inventory [ΔSTAI] Y-1, P=.37; ΔSTAI Y-2, P=.84), in-session face-to-face communication duration (excluding any presession AR technical preparation time; P=.73), or SEGUE scores (P=.60). Exploratory stratified analyses suggested larger comprehension gains with AR in the parietal and occipital approach strata. CONCLUSIONS: In a standardized preoperative SDM conversation, an integrated visualization-support package combining AR, patient-specific modeling, and interactivity improved neurosurgical decision makers' objective understanding and satisfaction without prolonging in-session, face-to-face communication duration. Larger multicenter trials with longer-term outcomes are warranted to confirm effectiveness and to evaluate implementation and cost considerations. TRIAL REGISTRATION: ISRCTN Registry ISRCTN11483487; http://www.isrctn.com/ISRCTN11483487.
20 Apr 2026
Read appraisal →PloS one
AI literacy among basic school teachers in Ghana: A structural equation modelling analysis
Understanding the synergy of Artificial Intelligence (AI) literacy among teachers is imperative for successful integration in educational settings. This study examined basic school teachers' (BSTs) AI literacy based on four key aspects: AI knowledge and understanding (KUAI), AI application (AAI), AI application evaluation (EAIA), and AI ethics (AIE). The study employed a cross-sectional survey design and purposive sampling to collect data from 319 BSTs in Ghana. A variance-based structural equation modelling (VB-SEM) approach was used to explore the relationships between these dimensions and provide insights into their interrelationships. The study found that AAI positively influenced KUAI, EAIA, and AIE. However, AIE did not significantly influence KUAI and EAAI. In addition, KUAI did not influence EAIA. The findings contribute to understanding the current state of AI literacy among BSTs in Ghana and highlight areas for improvement in AI education and professional development programmes.
19 Apr 2026
Read appraisal →JMIR cancer
Digital and Paper-Based Hospital Workflows and 60-Day Mortality in Acute Leukemia: Retrospective Natural Experiment
BACKGROUND: Electronic medical record (EMR) systems have been associated with better clinical workflows and fewer documentation errors. However, evidence regarding their effect on time-sensitive leukemia care in public hospitals in Latin America remains limited. OBJECTIVE: This study aimed to compare 60-day mortality and urgent supportive-care processes between 2 tertiary public hospitals in Mexico with different documentation models (integrated EMR vs traditional physical records [TPR]), under the hypothesis that digital workflows may facilitate more timely treatment. METHODS: We conducted a retrospective natural experiment including 274 patients with newly diagnosed acute leukemia treated between February 2023 and April 2025. Clinical characteristics, treatment intensity, complications during induction, antibiotic administration times, and survival outcomes were abstracted from finalized institutional records that are routinely reviewed at discharge by the institutional medical record committee as part of standard quality-assurance procedures. The primary outcome was 60-day mortality. Secondary outcomes included time to first antibiotic dose after recognition of febrile neutropenia, treatment-related complications, and the number of operational steps required for urgent care processes. A subgroup analysis of 70 patients with complete timing documentation was performed for antibiotic administration. Comparisons between hospitals were performed using univariate tests, Kaplan-Meier 60-day survival curves, and multivariable logistic regression. RESULTS: Of the 274 included patients, 104 (38%) were treated at the hospital using an integrated EMR, and 170 (62%) at the hospital using TPR. Sixty-day mortality was lower in the EMR hospital (6/104, 5.8%) than in the TPR hospital (61/170, 35.9%; P<.001). In the subgroup with complete timing data, the mean time from febrile episode recognition to first antibiotic administration was shorter in the hospital using EMR than in the hospital using TPR (54, SD 18.4 minutes; P<.001). Although the EMR hospital used a higher proportion of high-intensity regimens, patients in that hospital had better 60-day outcomes. In multivariable analysis, hospital type remained independently associated with 60-day mortality (odds ratio 0.11, 95% CI 0.05-0.26; P<.001), whereas kidney injury and hepatotoxicity were associated with worse outcomes. CONCLUSIONS: In this natural experiment, the hospital using an integrated EMR had a lower 60-day mortality and shorter time to antibiotic administration than the hospital using TPR. These findings are hypothesis-generating and suggest that digital workflows may contribute to more timely urgent supportive care, but they should be interpreted with caution, given the retrospective 2-center design and the potential for residual confounding.
19 Apr 2026
Read appraisal →Scientific reports
A hybrid dual-stream CNN framework with dynamic data augmentation and improved Manta Ray Foraging Optimization for robust glaucoma detection.
A progressive neurological condition, glaucoma is one of the main causes of irreversible blindness in the globe. Early detection is crucial to preventing irreversible vision loss however conventional diagnostic methods are often time-consuming, and heavily reliant on clinical expertise. This study presents an innovative deep learning framework designed to automate glaucoma detection while addressing key challenges such as data imbalance, image quality inconsistencies, and the need for accurate feature extraction. The proposed framework begins with a dedicated preprocessing pipeline that enhances image clarity, and isolates clinically relevant regions of interest. To handle class imbalance, a novel hybrid data augmentation strategy is introduced, combining geometric transformations with adaptive Gaussian noise injection that adjusts intensity according to image characteristics, thereby simulating realistic clinical variability. At its core, the framework leverages a dual-stream CNN architecture that integrates DenseNet121 for structural feature extraction and ResNet50 for texture representation. A lightweight channel-wise attention mechanism is then introduced to selectively emphasize clinically significant channels while suppressing redundant features, thereby balancing efficiency and interpretability. To optimize model performance while reducing computational overhead, an Improved Manta Ray Foraging Optimization (IMRFO) algorithm is employed. IMRFO enhances the standard MRFO with Partial Centroid Opposition-Based Learning (PCOBL) to dynamically fine-tune hyperparameters, including augmentation settings and transfer learning configurations. Experimental validation was conducted on four public benchmark datasets, ACRIMA, Drishti-Gs, ORIGA, and RIM-ONE-DL, demonstrating the framework's superior performance across all evaluation metrics. The model achieved 100.00% accuracy, precision, recall, and AUC on both ACRIMA and Drishti-Gs, with losses of 0.003 and 0.001, respectively. On ORIGA, it reached 99.70% accuracy, 99.80% precision, 99.30% recall, and 99.50% AUC (loss = 0.01). On RIM-ONE-DL, the model scored 99.90% accuracy, 99.70% precision, 99.50% recall, and 99.70% AUC (loss = 0.006). These findings confirm the framework's robustness and clinical applicability for effective glaucoma screening.
19 Apr 2026
Read appraisal →Journal of medical Internet research
Effectiveness of mHealth Interventions for Improving eHealth Literacy Among Patients With Chronic Diseases: Meta-Analysis and Systematic Review
BACKGROUND: With the widespread use of the internet and mobile devices, eHealth literacy promotion is critical for medical equity. Mobile health (mHealth) serves as a pivotal tool for enhancing eHealth literacy by providing accessible, interactive platforms for health information engagement. However, the evidence regarding the effectiveness of mHealth interventions on eHealth literacy among patients with chronic diseases remains inconclusive. OBJECTIVE: This study aimed to evaluate the effectiveness of mHealth interventions on eHealth literacy among patients with chronic diseases based on randomized controlled trials (RCTs) and summarize supportive evidence from quasi-experimental and qualitative studies. METHODS: A comprehensive search strategy was developed, and 8 electronic databases were systematically searched for studies published up to February 12, 2026. Patients with chronic diseases were included based on predefined inclusion criteria. The Cochrane risk of bias 2 tool for RCTs and the ROBINS-I tool for quasi-experimental studies were used to assess the risk of bias. Given the anticipated substantial heterogeneity among the studies included, we used a random-effects model based on the Hartung-Knapp-Sidik-Jonkman method to pool effect sizes. A narrative and quantitative synthesis of the findings was provided where appropriate. RESULTS: A total of 15 studies were included in this review, including 6 RCTs, 5 quasi-experimental studies, and 4 qualitative studies, involving a total of 2884 patients with chronic diseases. Meta-analyses of RCTs suggested that mHealth interventions could improve eHealth literacy, with a pooled mean effect size of standardized mean difference (SMD)=1. 19 (95% CI 0.14-2.23; P=.03; I²=97.75%; PI [prediction interval]=-2.68 to 5.05). Subgroup analyses by intervention targets showed that interventions on targets with specific disease produced larger mean effects (SMD=1.61; 95% CI 0.16-3.06; PI=-5.40 to 8.63), while interventions targeting the population with general chronic diseases produced smaller effects (SMD=0.36; 95% CI 0-0. 73; PI=-0. 21 to 0. 94). Analysis by intervention duration subgroup showed that the combined effect of studies with intervention duration <3 months was statistically significant (SMD=0.61; 95% CI 0.09-1.13; I²=88.04%; PI=-5.72 to 6.95); while the combined effect of studies with intervention duration ≥3 months was not statistically significant. Taking into account bias and the risk of GRADE (Grading of Recommendations, Assessment, Development, and Evaluation), the certainty of RCT evidence was moderate, and the certainty of quasi-experimental evidence was low. CONCLUSIONS: mHealth interventions could improve eHealth literacy among patients with chronic diseases on average. By using prediction intervals, this study reveals that the effectiveness of mHealth interventions is highly context-dependent and closely linked to implementation factors. Advancing beyond prior work, this study centers on eHealth literacy as a core outcome and integrates multiple types of evidence. Meanwhile, this finding emphasizes the need for evidence-based intervention programs and more rigorous implementation of intervention designs in future research.
19 Apr 2026
Read appraisal →PloS one
Evidence-based action plan for integrating artificial intelligence in an academic medical centre-a multidisciplinary approach
INTRODUCTION: As the effectiveness of artificial intelligence (AI) in enhancing various facets of healthcare delivery becomes more apparent, it is anticipated that AI will soon find its way into standard clinical practices, even in low and middle-income countries. The objective of this study was to create an action plan for integrating AI into medical education, research, and clinical practices utilizing the SWITCH model of change integrating both rational and emotional aspects. METHODS: This exploratory qualitative study employed reflexive thematic analysis of semi-structured interviews and a co-design workshop, followed by the collaborative development of an action plan. The study was conducted from May 2023 to May 2024, at the Aga Khan University, Karachi, Pakistan. The development of an action plan was informed by interviews, co-design workshop, and discussions with diverse group of academic leaders, healthcare professionals and medical students. All interviews, workshop sessions, and planning meetings were audio recorded, transcribed verbatim, and anonymized. Data management was conducted manually using Microsoft Word and Excel. Findings after thematic analysis of the qualitative interviews and findings from the co-design workshop were gradually, and inductively transformed into the content for the action plan for integrating AI into healthcare. RESULTS: The content analysis of interviews identified following four themes: AI opportunities, apprehensions toward AI-induced changes, pushing change through leadership styles, and importance of AI related capacity building. During the workshop, participants discussed aligning current AI knowledge with future requirements by identifying clear instructions, emotional motivators, and environmental changes required on the path of AI integration. The proposed action plan conceptualized AI integration as multidimensional change process comprising three domains: strategic actions, change pathway enablement and environmental readiness and human motivation operationalized through twenty actionable components. CONCLUSION: The study findings provide a context specific conceptual action plan for healthcare professionals to integrate AI into medical education, clinical service and research, Future work should focus on pilot implementation and empirical validations of the action plan across diverse healthcare settings to assess feasibility, effectiveness and scalability.
18 Apr 2026
Read appraisal →PloS one
Pretraining effective T5 generative models for clinical and biomedical applications
This paper presents a study of the impact of corpus selection and vocabulary design on the performance of T5-based language models in clinical and biomedical domains. We introduce five different T5-EHR models, each pretrained from scratch using different combinations of clinical and biomedical corpora alongside domain-specific vocabularies. We evaluated these models across a variety of clinical and biomedical tasks to quantify the impact of pretraining data and vocabulary tokenization choices on downstream performance. Our findings reveal the importance of aligning both pretraining corpus and vocabulary with the target domain. Models pretrained exclusively on clinical data achieve superior performance on clinical tasks, while adding biomedical data contributes only marginal gains in most cases, with a few exceptions. Similarly, the choice of vocabulary significantly influences model performance, with clinical-specific vocabularies outperforming general biomedical vocabularies in tasks requiring a deeper understanding of clinical language. Also, the T5 generative models perform competitively with state-of-the-art discriminative models on several biomedical benchmarks, demonstrating strong generalization to biomedical domain. Overall, these results emphasize that task-specific selection of corpus and vocabulary is essential for optimizing model performance in clinical and biomedical natural language processing (NLP).
18 Apr 2026
Read appraisal →PloS one
Operationalization of Severe Maternal Morbidity surveillance in administrative databases of the Brazilian public and private health systems: Brazil 2015-2022
The study of severe maternal morbidity (SMM) is an important strategy for improving the quality of obstetric care. The objective of this study is to assess SMM cases recorded in the Hospital Information System of the Unified Health System (SIH/SUS), which contains public funded hospitalizations, and in the database of the National Supplementary Health Agency (ANS), which contains hospitalizations paid for by health plans. This is a cross-sectional study using obstetric hospitalizations of women aged 10-49 years from both databases in the period 2015-2022. We assessed SMM cases using the World Health Organization's criteria for potentially life-threatening conditions (PLTC). Of the 26 PLTC criteria, 22 were operationalized through diagnoses during hospitalization, using International Classification of Diseases-10 codes, or medical procedures, using specific codes of each database. We estimated the frequency and odds ratio of death for each isolated criterion, for grouped criteria and for the total number of SMM cases, as well as the frequency of deaths according to the number of SMM criteria. We identified 18,807,757 obstetric hospitalizations in the SIH/SUS and 3,776,986 in the ANS database. The estimated proportion of SMM was 8.51% and 7.30% in SIH/SUS and ANS, respectively. Most cases presented only one criterion. Differences were observed among sectors, with a higher frequency of procedure-based criteria in the ANS database, compatible with the greater availability of resources in private services. The proportion of deaths increased with the increase in the number of criteria, and more severe criteria presented a higher odds ratio of death. The results indicate that it is possible to estimate the number of SMM cases in the country, both in public and private hospitalizations. Continued investment in the quality of recording in both databases are necessary so that the estimates can truly reflect the maternal morbidity of Brazilian women.
18 Apr 2026
Read appraisal →PloS one
EAC-Agent: A deep learning framework for multimodal emotion-aware conversational agent with contextual response generation
Conversational agents, commonly referred to as chatbots, have become an integral part of various important applications, such as customer support and virtual assistants in enterprise-level solutions. Despite their widespread use, current solutions predominantly rely on a single modality input, primarily text, which limits their ability to fully understand the correct emotions of users and respond accordingly. The increasing demand for multimodal inputs in conversation such as text, audio, and videos, highlights the need for a more comprehensive approach. Existing conversational agents face significant challenges in generating emotionally aware responses, as they lack the ability to effectively handle emotion embeddings, leading to limitations in emotional accuracy and contextual appropriateness. Focusing on these challenges, this research work leads to a novel multimodal approach which incorporates features from text, audio, and visual. The EAC-Agent proposed a sequence-to-sequence model with transformer, along with pre-trained embeddings such as GloVe. The self and cross-modal attention on text, audio and visual is used to generate a more emotionally intelligent response. EAC-Agent is validated through comparison with existing techniques on two benchmark datasets. The obtained results demonstrate superior performance in emotion classification and response generation. The proposed model achieves an accuracy of 76.27% on IEMOCAP and 67.57% on MELD for emotion recognition from multimodal user inputs. In addition, the emotion-aware response generation module shows clear improvements, with perplexity values of 39.01 and 42.30, BLEU scores of 0.31 and 0.30, and ROUGE-L scores of 0.45 and 0.44 on IEMOCAP and MELD, respectively. EAC-Agent demonstrates clear superiority over existing models and holds great promise for applications in customer service, healthcare, and other areas requiring empathetic and contextually appropriate interactions.
18 Apr 2026
Read appraisal →PloS one
A deep reinforcement based echo state network for network intrusion classification
Network intrusion classification referred to the process of monitoring and analyzing network traffic to identify suspicious activities or attacks. In this work, author proposed a novel approach to classify network intrusion by utilizing deep reinforcement learning (DRL), integrating a reservoir computing approach Echo State Network (ESN). A DRL-based approach improved upon traditional deep learning by adapting dynamically to novel/unknown and evolving attack patterns. Unlike static models, DRL continuously learned optimal strategies through interaction with the environment, allowing for better detection of previously unseen threats in real-time. To address the class imbalance often encountered in network intrusion datasets, we evaluated the performance of several advanced data balancing techniques, including Borderline-SMOTE, SMOTE-ENN, ADYSN, and K-means SMOTE. The findings demonstrated that the K-means-based data balancing method outperformed other techniques, resulting in the most robust performance across various metrics. Author conducted multi-dataset validation on benchmark datasets like NF-BoT-IoT, NF-UNSW-NB15, NF-ToN-IoT, NF-ToN-IoT-v2, NF-CSE-CIC-IDS2018 and NF-UNSW-NB15-v3 to ensure robustness across different network flow data. For adaptive modeling testing, author excluded some attack types from training data and included them in testing data (e.g., DoS, Backdoor attacks were excluded from the training data but included in the testing data (see Table 3)). The proposed approach enhanced the accuracy and reliability of intrusion detection, making it a viable solution for securing modern network infrastructures. The source code of this work is available in this Github repository (https://github.com/codewithkhurshed/DRLZDNIDS).
18 Apr 2026
Read appraisal →JMIR human factors
Interactive, Personalized Patient Decision Aid for COVID-19 Vaccination in Canada: User-Centered Design Approach
BACKGROUND: The COVID-19 pandemic highlighted the need for practical digital health tools to support informed decision-making amid rapidly evolving evidence and widespread misinformation. OBJECTIVE: We iteratively developed and refined VaxDA-C19, a bilingual (English and French) web-based patient decision aid designed to support informed decision-making in Canada about COVID-19 vaccination. VaxDA-C19 integrates interactive and personalized features aimed to enhance vaccine confidence, reduce cognitive overload, and respond to diverse informational needs. METHODS: VaxDA-C19 was developed using an iterative, user-centered design approach. Throughout the development process, we involved a citizen panel, health care professionals, user experience designers, and scientific experts to guide refinements. We also conducted usability testing sessions with adults in Canada, using semistructured interviews, comparative testing, and think-aloud protocols with thematic analysis. We ultimately conducted 4 design cycles in total with adults in Canada (users) and expert reviewers (experts). Cycle 1 involved 9 people (9 users), cycle 2 involved 22 people (22 users), cycle 3 involved 9 people (3 users and 6 experts), and cycle 4 involved 9 people (9 experts). RESULTS: In cycle 1, user feedback guided design decisions about how to present quantitative information and technical vaccine descriptions more simply. In cycle 2, while most users (9/11, 82%) favored in-depth explanations of vaccine development, a few raised concerns about content that could be perceived as politically charged. Cycle 3 identified usability improvements, including more explicit navigation controls, simplified medical terminology, and optimized interactive components (avatars and sliders). Expert reviews in cycle 4 refined linguistic consistency, mobile responsiveness, content transparency, and scientific accuracy, emphasizing explicit instructional guidance and bilingual accessibility. CONCLUSIONS: Our iterative process produced a personalized, bilingual digital decision aid to support evidence-informed, values-congruent decisions about COVID-19 vaccination. A randomized controlled trial will further evaluate VaxDA-C19's impact on vaccination intentions, knowledge retention, emotional responses, decisional conflict, and decisional regret. If it proves effective, the patient decision aid may also be used as a platform to support other vaccine decisions, namely, influenza, measles, shingles, pertussis, and potentially other emerging infectious diseases.
18 Apr 2026
Read appraisal →International journal of cardiology
Predicting functional results of percutaneous coronary intervention using machine learning modelling
BACKGROUND: Post-percutaneous coronary intervention (PCI) Murray's law-based quantitative flow ratio (μFR) is associated with long-term clinical outcomes. A tool capable of accurately predicting post-PCI μFR before intervention could support procedural planning, reduce the risk of suboptimal physiological results, and improve prognosis. OBJECTIVES: To develop and validate machine-learning models that predict continuous post-PCI μFR using only pre-procedural angiographic, physiological and clinical data, and to assess their ability to classify PCI outcomes as optimal (μFR ≥ 0.91) or sub-optimal (μFR < 0.91). METHODS: Four machine learning models were trained using pre-PCI variables. Internal bootstrap validation (1000 iterations) identified the best-performing model based on lowest root mean square error (RMSE) for continuous prediction. Predicted μFR values were subsequently used to classify PCI outcomes. RESULTS: In 343 vessels (291 patients), machine learning achieved high accuracy for continuous post-PCI μFR prediction (RMSE 0.036; 95% CI: 0.033-0.040; mean absolute error 0.030; 95% CI: 0.027-0.032; mean absolute percentage error 3.2%; 95% CI: 2.9-3.5), indicating reliable estimation of post-PCI μFR using only pre-procedural data. When the predicted μFR was used to classify PCI outcomes, performance remained clinically meaningful, with accuracy 0.72 (95% CI: 0.70-0.75), area under the curve 0.72 (95% CI: 0.69-0.74), sensitivity 0.90 (95% CI: 0.88-0.93), and specificity 0.29 (95% CI: 0.23-0.34). The high sensitivity ensures reliable upfront identification of vessels likely to achieve optimal physiology. CONCLUSIONS: Machine-learning models accurately predict post-PCI μFR and reliably distinguish optimal from sub-optimal outcomes before intervention. This approach supports personalized PCI planning and improves strategy selection.
17 Apr 2026
Read appraisal →Journal of medical Internet research
Artificial Intelligence for Identifying Patient-Reported Outcome and Experience Measures in Oncology: Retrospective Cross-Sectional Study Using ClinicalTrials.gov
BACKGROUND: Traditional health care systems have evolved to increasingly recognize patients' perspectives as key to improving the quality of care, especially in oncology. Hence, patient-reported outcome measures (PROMs) and patient-reported experience measures (PREMs) can enhance patient-health care provider communication while facilitating individualized care. This tailored approach improves patient outcomes and highlights the need for research methods to account for variability in patient experiences across diverse sociodemographic and clinical backgrounds while proposing artificial intelligence (AI) to automate and accelerate the identification of PROMs and PREMs. OBJECTIVE: This study aimed to estimate the proportion of clinical studies including PROMs or PREMs, using either a traditional expert-based identification method or an AI-enriched approach. METHODS: In a retrospective cross-sectional study using the ClinicalTrial.gov database, we focused on oncology studies between 2012 and 2022. Two methods were used to identify PROMs and PREMs: (1) a traditional expert-based method, where an algorithm identified PROMs and PREMs from a list of 346 oncology-specific PROMs and PREMs (extracted from the PROQOLID database, Mapi Research Trust) and/or 11 PROM- and PREM-specific terms, and (2) an AI-enriched method using a bidirectional encoder representations from transformers model, trained on 2399 outcomes labeled by experts. To evaluate algorithm performance, results were compared with expert decisions. Studies were classified by PROM/PREM use, analyzed using logistic regression to identify drivers, and a named entity recognition model identified frequently used measures. RESULTS: A total of 24,491 studies were included. According to the traditional expert-based algorithm, 7549 (31%) studies used at least one PROM or PREM, as compared to 8029 (33%) studies identified by the AI-enriched algorithm, increasing from 2012 to 2022 (P<.001). With 90% (95% CI 88%-92%) accuracy, the AI-enriched algorithm outperformed (P<.001) the expert-based algorithm (84%, 95% CI 82%-86%) in identifying PROMs and PREMs. Breast and digestive cancers accounted for nearly 50% of all oncology studies using PROMs and PREMs, with the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30 being the most frequently used. As expected, trials in phases 2 to 4 more frequently included PROMs and PREMs than preclinical or early phase 1 studies (odds ratio [OR] 1.8, 95% CI 1.1-2.8 for phase 2; OR 3.6, 95% CI 2.3-5.8 for phase 3; and OR 2.6, 95% CI 1.6-4.4 for phase 4). In observational studies, cross-sectional and prospective studies incorporated more PROMs and PREMs than retrospective studies (OR 4.6, 95% CI 3.3-6.4 and OR 3.2, 95% CI 2.5-4.1, respectively). CONCLUSIONS: Our study shows that an AI-enriched algorithm outperforms traditional expert methods in identifying PROMs and PREMs in oncology. Combining expert-labeled data with AI enables scalable, automated trial monitoring, supports efficient research, informs stakeholders, enhances patient-centered decisions, and can be extended to other diseases and databases.
17 Apr 2026
Read appraisal →JMIR cardio
Large Language Models in Cardiology: Systematic Review
BACKGROUND: Large language models (LLMs) are increasingly used in health care, but their role in cardiology has not yet been systematically evaluated. OBJECTIVE: This review aimed to assess the applications, performance, and limitations of LLMs across diverse cardiology tasks, including chronic and progressive conditions, acute events, education, and diagnostic testing. METHODS: A systematic search was conducted in PubMed and Scopus for studies published up to April 14, 2024, using keywords related to LLMs and cardiology. Studies evaluating LLM outputs in cardiology-related tasks were included. Data were extracted across 5 predefined domains and the risk of bias was assessed using an adapted QUADAS-2 tool (developed by Whiting et al at the University of Bristol). The review protocol was registered in PROSPERO (CRD42024556397). RESULTS: A total of 33 studies contributed quantitative outcome data to a descriptive synthesis. Across chronic conditions, ChatGPT-3.5 (OpenAI) answered 91% (43/47) heart failure questions accurately, although readability often required college-level comprehension. In acute scenarios, Bing Chat omitted key myocardial infarction first aid steps in 25% (5/20) to 45% (9/20) of cases, while cardiac arrest information was rated highly (mean 4.3/5, SD 0.7) but written above recommended reading levels. In physician education tasks, ChatGPT-4 (OpenAI) demonstrated higher accuracy than ChatGPT-3.5, improving from 38% (33/88) to 66% (58/88). In patient education studies, ChatGPT-4 provided scientifically adequate explanations (5.0-6.0/7) comparable to hospital materials but at higher reading levels (11th vs 7th grade). In diagnostic testing, ChatGPT-4 interpreted 91% (36/40) electrocardiogram vignettes correctly, significantly better than emergency physicians (31/40, 77%; P< .001), but showed lower performance in echocardiography. CONCLUSIONS: LLMs show meaningful potential in cardiology, especially for education and electrocardiogram interpretation, but performance varies across clinical tasks. Limitations in emergency guidance and readability, as well as small in silico study designs, highlight the need for multimodal models and prospective validation.
17 Apr 2026
Read appraisal →JMIR formative research
A Survey of Primary Care Clinician Experiences With Electronic Health Record-Based Clinical Decision Support to Improve HIV Pre-Exposure Prophylaxis Prescribing
A survey of primary care clinicians suggests that a clinical decision support tool to support sexual risk assessment and prescribing of HIV pre-exposure prophylaxis was appropriate and useful for identifying at-risk patients, but uptake was hindered by workflow and usability barriers, which underscores the importance of postimplementation clinician feedback to improve the use of clinical decision support tools.
17 Apr 2026
Read appraisal →Philosophical transactions of the Royal Society of London. Series B, Biological sciences
Crowdsourced versus large language models forecasting: evidence for the accuracy-correlation effect
Over the past quarter century, crowdsourced forecasting has largely outperformed individual forecasters. Today, large language models (LLMs), aggregating human knowledge at scale, constitute a new form of collective intelligence (CI). A central question is how LLM predictive accuracy associates with human-AI correlation, and whether this relationship exceeds what would be expected if both merely track the same underlying truth. We investigate this through the accuracy-correlation effect (ACE), which posits that as algorithmic systems improve, they increasingly correlate with human predictions, potentially diminishing human value in hybrid ensembles. Using 76 model × prompt forecast sets from 16 LLMs on 580 resolved ForecastBench questions, we computed LLM accuracy and correlations with two human aggregates (superforecasters, general public), separately for databases (n = 526) and prediction markets (n = 54) questions. Linear mixed-effects models show a robust positive association between LLM accuracy and human-AI correlation that substantially exceeds independent-errors predictions. Correlations were lower for superforecasters than for the general public, and weaker for markets than for data questions. These results support ACE while indicating that increasing correlation reflects more than improved signal tracking alone, suggesting that simultaneous increases in accuracy and correlation may reduce optimal human weights in data-rich settings, while human judgement retains critical value in contextual reasoning scenarios. This article is part of the theme issue 'The evolution of collective intelligence'.
17 Apr 2026
Read appraisal →Journal of medical Internet research
Beyond Missingness: Systematizing Methods for Comprehensive Data Fitness Assessment in Clinical Research
Secondary use of clinical data offers unprecedented opportunities to rapidly conduct large-scale research and improve patient care. However, incomplete understanding of data quality requirements for a study often causes significant delays in executing analyses and validating results. Current practice has largely followed 2 paths. First, multi-institutional networks have developed general data quality programs, but these are typically tied to unique network characteristics and do not address study-specific requirements well. Second, models have been proposed to formalize the requirements for data fitness analyses without extending to the methods needed to meet these requirements. More recently, tools have been developed to conduct cohort-centric screening, focusing on generally applicable structural checks such as missingness or facial implausibility. These provide a first level of information but incompletely capture the fitness requirements of an analysis. In turn, investigators conduct per-study exploratory analyses, but these efforts are typically ad hoc and partially reported, which can hinder reproducible science and delay advances in patient care. Analogously to advances over the past decade in data modeling and reproducible analytics, there is a need for a more systematic, capable approach to study-specific data quality assessment (SSDQA). We discuss such a model, which guides improved SSDQA design and implementation, including metadata for consistent annotation and reporting of data quality assessment results. The model integrates theoretical principles of data quality testing with pragmatic considerations of application to clinical data, providing a consistent approach to specifying data quality assessment checks. Additionally, it proposes to regularize check application through a standard set of options. The SSDQA model builds on current practice, providing a path toward more complete, sound, and reproducible assessments. These characteristics foster multidisciplinary collaboration to identify data quality issues that, in turn, inform decisions about study design and provide important context that has a bearing on adoption of results.
17 Apr 2026
Read appraisal →International journal of medical informatics
Rule-augmented constraint learning for semantic error detection in MIMIC-III knowledge graph
High-quality, error-free data is essential for developing reliable data-driven models, particularly in clinical decision support systems where inaccurate predictions can have serious consequences. While KGs offer a structured and semantically rich representation for clinical data, ensuring their consistency and correctness remains a challenge. Existing rule mining techniques provide solutions for the automatic extraction of logical constraints from KGs, but they often produce redundant or clinically irrelevant rules, especially when dealing with numeric or categorical literals such as age or lab values. KG constraints-rules intended to capture implausible or conflicting facts in the KG-can be used to spot semantic errors: facts that might conform to the underlying schema but contradict domain knowledge. In this work, we propose a novel framework for constraint learning in clinical KGs that identifies and transforms high-confidence rules into clinically plausible constraints. We propose two approaches, based on class disjointness and literal clustering combined with rule mining. We validate the clinical relevance of these generated rules using expert-curated constraints and large language models (LLMs). The results on the MIMIC-III clinical dataset show that rule filtering based constraint learning effectively preserves clinically meaningful rules that align with established medical knowledge. For numeric data, we achieve reliable value groupings through our clustering-based method, and the rules derived from these groupings were validated by LLMs. Their outputs confirm the clinical relevance of a portion of those discovered rules. By providing interpretable and scalable solutions to semantic inconsistencies in KGs, this study contributes to increasing the KG trustworthiness and its clinical usability.
17 Apr 2026
Read appraisal →JMIR cancer
Exploring the Adoption of Mobile Health Apps Among Patients with Head and Neck Cancer After the COVID-19 Pandemic: Secondary Analysis of Cross-Sectional Survey
BACKGROUND: Patients with head and neck cancer (HNC) face significant challenges in accessing coordinated care due to the complexity and multimodality of their treatment and the impact on vital functions. The COVID-19 pandemic has disrupted cancer care while accelerating the adoption of digital health solutions. Mobile health (mHealth) apps offer potential solutions for remote symptom monitoring, communication between patients and providers, and continuity of care. Nevertheless, their acceptance among patients with HNC remains limited due to age-related digital divides and concerns about accessibility. OBJECTIVE: The aim of this study was to investigate the impact of the COVID-19 pandemic on patients with HNC and to explore their attitudes toward mHealth apps as a supplement to cancer treatment. METHODS: A secondary analysis of a cross-sectional survey was conducted between January 2023 and May 2024 at Julius Maximilian University of Würzburg in Germany. A total of 355 patients with HNC were recruited and completed the structured "Cancer and COVID-19" survey via the Corona Health app or in paper form. The 25-question survey assessed sociodemographic information, the impact of the pandemic on diagnosis/treatment/aftercare, and interest in mHealth apps. Descriptive statistics and bivariate correlations were used for the analysis. RESULTS: The cohort comprised 261 (74%) men and 94 (26%) women with an average age of 67 (SD 11.5) years. Most participants (n=264, 74%) stated that the pandemic had no impact on their cancer treatment, although 20% (n=71) experienced disruptions, particularly in follow-up appointments and treatment monitoring. Only 10% (n=36) currently used health apps, but 57% (n=203) expressed a willingness to use mHealth technologies. Younger patients, patients with higher education, and participants who were more affected by the pandemic showed greater openness to digital health solutions. The most significant barriers included age, digital literacy, and perceived usefulness, while preferred app features included interaction with physicians (n=160, 45%) and data sharing with researchers (n=153, 43%). CONCLUSIONS: Although the COVID-19 pandemic had only a limited direct impact on HNC care at the institution at hand, it revealed significant patient interest in mHealth apps. However, significant barriers remain, particularly among older adults with lower digital literacy. Future mHealth initiatives should focus on improving digital literacy, addressing privacy concerns, demonstrating clinical benefits, and developing personalized, accessible solutions to optimize cancer care for this vulnerable population.
17 Apr 2026
Read appraisal →JMIR medical informatics
Research on Risk Transfer Pathways for Lung Cancer Among Middle-Aged and Older Individuals Using Deep Reinforcement Learning: Retrospective Cohort Study
BACKGROUND: Nowadays, lung cancer has a significantly high incidence rate worldwide. The mortality rate of lung cancer continues to rise; it is more common in middle-aged and older individuals and poses a great threat to human health. OBJECTIVE: This study aimed to assess the lung cancer risk among middle-aged and older individuals in a timely manner and to establish an efficient pathway for the risk transfer. METHODS: We proposed a deep reinforcement learning model based on deep Q-network (DQN) to explore the risk transfer pathway for lung cancer among middle-aged and older individuals. Risk stratification of lung cancer occurrence was deduced through deep neural network. The DQN model was developed using the Health and Retirement Study cohort for model training and internal validation. We also used the China Health and Retirement Longitudinal Study cohort for model external validation. Transfer simulation of multiple pathways in different cycles was calculated in a stratified risk groups-leveraged DQN model. RESULTS: We developed and evaluated the DQN method for optimizing the risk transfer pathway among middle-aged and older individuals, with accuracy ranging from 0.917 (95% CI 0.896-0.928) to 0.949 (95% CI 0.909-0.961) and area under curve ranging from 0.906 (95% CI 0.887-0.933) to 0.927 (95% CI 0.893-0.938). External validation was conducted to assess the model's effectiveness and availability. A total of 8780 and 3763 samples from the Health and Retirement Study were used for model training and internal testing, respectively, and 16,442 samples from the China Health and Retirement Longitudinal Study were used for external validation. The results showed that DQN models illuminated the optimal risk transfer pathways for stratified risk groups. Lung cancer incidence in the high risk group had declined by 68.2% through risk transfer of model-based simulation, which had declined by 56.9% in the medium risk group. Through simulative intervention and risk transition deduced from the DQN model, lung cancer incidences of the high risk, medium risk, and low risk groups were obviously decreased. CONCLUSIONS: A DQN-based deep reinforcement learning model was proposed and validated to develop and simulate a risk transfer pathway of lung cancer among middle-aged and older individuals. Risk stratification supplied an effective foundation for lung cancer risk transition.
17 Apr 2026
Read appraisal →JMIR human factors
Comparing Images of Depression in Mass Media and AI-Generated Pictures: Mixed Methods Study
BACKGROUND: Images play an important role in reducing stigma related to mental health, which often is distorted in the media. In recent years, generative artificial intelligence (AI) has been used to generate images related to mental health. However, first reports suggest that AI-generated images do not depict mental health conditions accurately. In-depth studies on the topic of mental health representations in AI-generated images are still missing. OBJECTIVE: The main objective of this study is to analyze and compare the visual representation of depression in mass media and in AI-generated images. METHODS: The methodologies used were discussion groups (15 participants) and a quasi-experimental online survey (792 interviewees), aimed at people with depression and young people. RESULTS: The results showed that both the images used in the media and those generated by AI reproduced stereotypes and stigmas about depression. However, participants considered AI-generated pictures to be more stereotypical, stigmatizing, and more likely to have a negative impact on people with depression. In contrast, media images were considered more appropriate, realistic, inclusive, and that better reflected the relationship between gender and depression. Statistically significant differences were observed between the control and test groups in both people with depression and young people (P<.001), indicating that when people were aware of what images were AI-generated, they tended to reject them to a greater extent. CONCLUSIONS: Considering the current trend toward the widespread use of AI in mental health communication, it is crucial to promote closer collaborations between science journalists, AI developers, and mental health experts, including patients' associations, as well as a shift toward user-participatory AI design.
17 Apr 2026
Read appraisal →JMIR formative research
Development of a Contextualized, Research-Based Flemish Assessment Framework for Digital Care, Assistance, and Support: Delphi Study
BACKGROUND: The rapid evolution of digital technologies has transformed health, mental health, and social care, offering new modalities of digital care, assistance, and support through web-based platforms, mobile apps, extended reality, wearables, and artificial intelligence systems. Despite this proliferation, there is little consensus on what constitutes "high-quality" digital care. Challenges persist regarding data security, interoperability, accessibility, sustainability, and professional competence, whereas existing standards and regulations provide fragmented guidance. OBJECTIVE: This study aimed to develop a contextualized, consensus-based quality assessment framework for digital care, assistance, and support in Flanders, Belgium. For this purpose, perspectives across technology, organizational processes, and professional competencies were integrated. METHODS: The study used a multiphase design comprising (1) 10 expert interviews with Flemish government officials; (2) a narrative literature review of 303 peer-reviewed and gray literature sources; (3) a 3-round Delphi study with 50 experts across 5 domains (end users, facilitators, technology developers, deontology and ethics experts, and digital inclusion and media literacy experts); and (4) 4 complementary focus groups and 3 interviews with specialists in artificial intelligence, regulation, social work, mental health, and IT. The Delphi rounds gathered iterative feedback through open-ended elicitation, structured rating, and classification of quality criteria. Quantitative data were analyzed using descriptive statistics, whereas qualitative feedback was subjected to thematic analysis. RESULTS: A total of 50 experts participated in round 1, a total of 40 (80%) participated in round 2, and 27 (54%) participated in round 3. Round 1 generated 577 unique quality criteria, consolidated into 26 clusters organized under 3 pillars: technology, organization, and professional competencies. The relative importance across pillars was balanced (mean score 37.29, SD 12.38 for technology; 33.33, SD 10.39 for professional competencies; and 29.80, SD 10.45 for organizations). Accessibility, reliability, and safety ranked highest for the technology; vision, quality monitoring, and infrastructure ranked highest for organization; and support, digital competencies, and ethics ranked highest for professional competencies. The finalized framework included 112 criteria, of which 35 (31.3%) were designated as optional and 77 (68.8%) were designated as minimum requirements. Focus groups and interviews validated the framework's comprehensiveness and usability, emphasizing proportional implementation, user centrality, and alignment with European Union regulations. Stakeholders highlighted the need for tools, training, and governance mechanisms to ensure adoption and sustainability. CONCLUSIONS: This study produced a codeveloped, context-sensitive quality assessment framework that balances technological robustness, organizational readiness, and professional competence in digital care, assistance, and support. The framework can serve both as a quality safeguard and a developmental road map. Accompanying self-assessment and governance tools enhance practical applicability. Implementation success will depend on governmental support, resource allocation, and structured feedback loops. Future research should pilot the framework in real-world settings, assess its impact, and establish mechanisms for continuous updates to maintain relevance in a rapidly evolving digital landscape.
17 Apr 2026
Read appraisal →JMIR formative research
Fact-Checking Large Language Model Responses to a Health Care Prompt: Comparative Study
BACKGROUND: Large language models use machine learning to produce natural language. These models have a range of potential applications in health care, such as patient education and diagnosis. However, evaluations of large language models in health care are still scarce. OBJECTIVE: This study aimed to (1) evaluate the accuracy and efficiency of automated fact-checking by 2 large language models and (2) illustrate a process through which a large language model might support a patient in redrafting a prompt to include key information needed for patient safety. METHODS: A parallel comparison of 2 large language models and 3 human experts was conducted. A clinical scenario was devised in which a woman aged 23 years questions the safety of retinoid treatment for acne by sending prompts to 2 large language models (GPT-4o and OpenBioLLM-70B). GPT-4o and OpenBioLLM-70B were asked to suggest improvements to the patient's initial prompt to elicit key information for clinical decision-making. After the patient sent the revised prompt to the large language models, the models were then asked to fact-check the final response. To test the generalizability of automated fact-checking, a set of 20 clinical statements on disparate topics, mostly related to drug indications, contraindications, and side effects, was developed. The large language models also fact-checked these 20 medical statements. The results were compared against the evaluations of 3 clinical experts. The outcome measures were as follows: (1) percentage of accuracy of automated fact-checking, (2) time to complete fact-checking, and (3) a binary outcome for prompt redrafting (advising the patient to revise her prompt by naming her acne medication to address safety concerns). RESULTS: For the scenario of a patient with acne, GPT-4o and OpenBioLLM-70B both had 86% agreement with the clinical experts' fact-checking. The large language models did not consistently convey the urgency of discontinuing isotretinoin treatment when pregnancy is suspected. In addition, the models did not adequately convey the importance of folic acid supplementation during pregnancy. For the set of 20 medical claims, GPT-4o fact-checking had 100% agreement with that of human experts, whereas OpenBioLLM-70B had 95% agreement. OpenBioLLM-70B diverged from human experts and GPT-4o on 1 question related to pediatric use of antihistamines. The expert fact-checks took a mean time of 18 (SD 3.74) minutes, GPT-4o took 42 seconds, and OpenBioLLM-70B took 33 minutes. The GPT-4o responses for the acne scenario had some inconsistencies but zero fabrication and no obvious omissions. In contrast, OpenBioLLM-70B omitted 1 key information item needed for patient safety. CONCLUSIONS: GPT-4o can interact with patients to improve the quality and comprehensiveness of the information contained in health-related prompts. GPT-4o and OpenBioLLM-70B can conduct efficient fact-checking that is close to the level of accuracy of human experts. Human experts need to perform additional checks for accuracy and safety.
17 Apr 2026
Read appraisal →A&A practice
Automated Identification of Cardiopulmonary Disease Cases for Preoperative Risk Stratification Using Machine Learning: A Retrospective Analysis
BACKGROUND: Preoperative chart review is time-consuming and prone to errors, particularly for cardiopulmonary conditions that impact anesthetic planning. We developed a guideline-aligned "clinical insight bot" that mines free-text documentation to surface perioperative cardiovascular risk signals relevant to the 2024 Mult Society perioperative guideline for noncardiac surgery. METHODS: We analyzed 1000 de-identified medical cases from the PhysioNet MIMIC database. Medical terminology was extracted using regex-based NLP and categorized into 13 clinical specialties. Text features were encoded using TF-IDF vectorization and 1536-dimensional semantic embeddings stored in a PostgreSQL vector database (pgvector). Four machine learning models-Logistic Regression, Random Forest, Support Vector Machine (SVM), and Naive Bayes-were trained with stratified fivefold cross-validation to classify cases as "cardiopulmonary-only" versus "mixed/other." Performance was evaluated using accuracy, precision, recall, and F1 score, with statistical comparison via McNemar's test and bootstrap confidence intervals. RESULTS: In a held-out test set of 200 notes (28 positive; 172 negatives; ~14% prevalence), a linear support vector machine achieved the best overall balance (F1 ≈ 0.71), with high precision (positive predictive value 0.94) and very low false positive rate (FPR) (1/172 ≈ 0.6%). False negatives were the dominant residual error class. The pipeline processed documents near-instantaneously and, when scaled to 1000 notes, replaced on the order of tens of clinician review hours (≈100× efficiency gain) while maintaining performance across common preoperative document types. CONCLUSIONS: A lightweight, guideline-aligned insight bot can transform unstructured preoperative notes into concise, stepwise prompts that flag cardiovascular risk signals before the day of surgery. High precision with a very low FPR supports safe integration with anesthesiology workflows by minimizing paging noise, whereas time savings create operational and financial value. Future work should emphasize multicenter validation, structured data fusion (including labs, imaging, and vitals) to improve sensitivity, and prospective evaluation of downstream clinical and operational outcomes.
17 Apr 2026
Read appraisal →JMIR mental health
Determinants of Digital Health Literacy Among Patients With Serious Mental Illness: Cross-Sectional Survey
BACKGROUND: Individuals with serious mental illness increasingly use digital devices and the internet to access health information and services but often face challenges when navigating digital tools, which may limit the benefits they receive from online health resources and digital health care services. OBJECTIVE: The objective of our study was to assess digital health literacy among individuals with serious mental illness and identify factors influencing this literacy. METHODS: Participants were recruited, using convenience sampling, from 2 psychiatric clinics, 1 day-care center, and 4 halfway houses in Taipei, Taiwan, between May 2024 and February 2025. Self-reported data were collected using a survey that incorporated the eHealth Literacy Scale, the Attitudes Toward Computer/Internet Questionnaire, and the Mobile Device Proficiency Questionnaire. Generalized linear modeling was applied to identify factors associated with digital health literacy. RESULTS: Among 255 participants included in the analysis, 83.5% (n=213) reported owning at least 1 digital device. Digital health literacy was significantly lower among individuals who reported greater perceived difficulty in using digital tools (B=-1.533, 95% CI -2.350 to -0.717; P<.001) and higher distrust in online information (B=-0.986, 95% CI -1.916 to -0.056; P=.04). By contrast, greater mobile device proficiency (B=0.144, 95% CI 0.008-0.281; P=.04) and self-efficacy (B=1.777, 95% CI 0.376-3.177; P=.01) were positively associated with digital health literacy. CONCLUSIONS: Despite widespread device ownership, digital health literacy was varied and generally suboptimal among patients with serious mental illness. Perceived difficulty and distrust emerged as major barriers; proficiency and self-efficacy facilitated higher literacy. These findings highlight the need for mental health professionals to integrate tailored digital skills training, confidence-building strategies, and ongoing support into digital health interventions for individuals with serious mental illnesses.
16 Apr 2026
Read appraisal →JMIR research protocols
Integrating Gait Analysis and AI Into Knee Osteoarthritis Care: Protocol for a 3-Phase Participatory Qualitative Study
BACKGROUND: Knee osteoarthritis (OA) leads to pain, disability, and reduced quality of life. For advanced stages, knee arthroplasty surgery is the standard treatment, yet dissatisfaction and persistent mobility deficits remain common. Current surgical decision-making processes seldom incorporate objective predictors of outcomes, such as biomechanical data. Advances in computer vision, wearable sensors, and artificial intelligence now enable efficient capture and interpretation of clinically prognostic gait features in real-world settings. However, the clinical adoption of such innovations remains limited, hindered by usability challenges, misalignment with stakeholder needs, and system-level barriers. OBJECTIVE: This study aims to inform the development of a knee OA clinical decision support (CDS) tool that integrates gait analysis and artificial intelligence with clinical decision-making. More broadly, it seeks to advance a framework for participatory digital health CDS tool implementation by identifying and addressing suboptimal workflows, stakeholder priorities, and organizational challenges. METHODS: This study uses a 3-phase participatory design process, with in-depth interviews conducted in each phase to generate insights that inform subsequent stages. Phase 1 investigates perceived barriers to and facilitators of the implementation of a digital CDS tool and examines current workflow processes in knee OA management. Phase 2 focuses on identifying and defining user requirements to guide solution ideation for translational digital decision support. Phase 3 assesses user acceptance of prototyped solutions. Field observations in phase 1 and usability testing in phase 3 will supplement interview findings with real-world evidence. RESULTS: The study was funded in November 2024. Ethics approval was granted by the Nova Scotia Health Research Ethics Board on July 28, 2025. Data collection began in September 2025. As of February 2026, a total of 6 clinicians have been interviewed, with all 6 interviews transcribed and 4 coded as part of phase 1 analysis. Study completion is anticipated by August 2026. Findings will be published by January 2027. CONCLUSIONS: Our stakeholder-driven approach prioritizes both technological rigor and practical usability, addressing key human factors for successful implementation. While initially focused on Nova Scotia's orthopedic setting, this research provides a scalable framework for integrating digital decision support tools into broader clinical environments, advancing innovation in knee OA care and beyond.
16 Apr 2026
Read appraisal →JMIR nursing
Application of the Technology Acceptance Model to Predict Nursing Students' Intention to Use Informatics: Cross-Sectional Study
BACKGROUND: Nursing informatics is essential for digital health transformation; however, the technology acceptance of undergraduate nursing students in Saudi Arabia remains underexplored. OBJECTIVE: This study examined factors influencing nursing students' intention to use informatics technologies using the technology acceptance model. METHODS: A cross-sectional survey was conducted with 132 undergraduate nursing students. Data were analyzed using descriptive, correlational, and hierarchical regression analyses. RESULTS: Perceived usefulness (mean 3.68, SD 1.22) and perceived ease of use (mean 3.64, SD 1.32) were the strongest predictors of acceptance, together explaining 87% of the variance (R²=0.87; β=0.323 for usefulness, P<.001; β=0.195 for ease of use, P=.032). Only 25.8% (n=34) of the students often used electronic health records, while 31.8% (n=42) had no electronic health record experience, indicating a clear gap in practical informatics exposure. CONCLUSIONS: Nursing students' acceptance of informatics is primarily driven by its perceived usefulness and perceived ease of use. These findings highlight the urgent need to integrate practical, user-centered informatics training and clinical simulation into undergraduate nursing curricula to better prepare students for technology-based practice.
16 Apr 2026
Read appraisal →BMJ open
Generative artificial intelligence-driven chatbots and medical misinformation: an accuracy, referencing and readability audit
OBJECTIVES: Artificial intelligence (AI)-driven chatbots have been rapidly adopted across research, education, business, marketing and medicine. Most interactions, however, come from non-experts using chatbots like search engines, including for everyday health and medical queries. DESIGN: We conducted an original study to audit chatbot responses in health and medical fields prone to misinformation. METHODS: Five popular chatbots were assessed: Gemini (Google), DeepSeek (High-Flyer), Meta AI (Meta), ChatGPT (OpenAI) and Grok (xAI). In February 2025, each chatbot was prompted with 10 questions from five categories: cancer, vaccines, stem cells, nutrition and athletic performance. We deployed an adversarial-like framework, using open- and closed-ended prompts designed to strain models toward misinformation or contraindicated advice. Two experts from each category rated responses as 'non-problematic', 'somewhat problematic' or 'highly problematic' using a coding matrix based on objective, predefined criteria. Citations were scored for accuracy and completeness, and each response was given a Flesch Reading Ease score. RESULTS: Nearly half (49.6%) of responses were problematic: 30% somewhat problematic and 19.6% highly problematic. Response quality did not differ significantly among chatbots (p=0.566) but Grok generated significantly more highly problematic responses than would be expected under a random distribution (z-score +2.07, p=0.038). Performance was strongest in vaccines (mean z-score -2.57) and cancer (-2.12), and weakest in stem cells (+1.25), athletic performance (+3.74) and nutrition (+4.35). Chatbot outputs were consistently expressed with confidence and certainty; from 250 total questions, there were only two refusals to answer (0.8%), both from Meta AI. Reference quality was poor, with a median completeness score of 40% (Q1-Q3: 20-67%). Chatbot hallucinations and fabricated citations precluded any chatbot from producing a fully accurate reference list. All readability scores were graded as 'Difficult' (30-50), equivalent to college sophomore-senior level. CONCLUSIONS: The audited chatbots performed poorly when answering questions in misinformation-prone health and medical fields. Continued deployment without public education and oversight risks amplifying misinformation.
16 Apr 2026
Read appraisal →International journal of medical informatics
Prioritising digital health technologies in Australian community pharmacy: a delphi study identifying barriers, enablers, and policy implications for implementation
OBJECTIVE: To determine priority digital health technologies for Australian community pharmacies and identify the key barriers, enablers, and policy/funding factors that can inform future implementation planning through expert consensus. METHODS: A two-round Delphi study was conducted with 31 experts representing pharmacy, academia, policy, and digital health. In Round 1, participants identified priority technologies, barriers, and enablers. In Round 2, 27 participants ranked five technologies, nine policy options, and six financial models. Consensus was assessed using descriptive statistics and interquartile ranges (IQRs). RESULTS: E-prescriptions and My Health Record (MyHR) were ranked as top priorities (mean = 1.70 and 2.22; IQR ≤ 1.0). Key barriers included financial constraints, interoperability issues, and digital literacy gaps. Telehealth incentives received the strongest agreement among participants, while reimbursement-based funding and government support were rated as the most supportive financial models for implementation. Broader enablers, such as a national medicine repository and stronger cross-disciplinary collaboration, were also endorsed. CONCLUSION: Digital health adoption in community pharmacy requires prioritisation of core technologies, improved system integration, workforce training, and practical funding mechanisms. These findings offer guidance for policymakers, pharmacy leaders, and digital health stakeholders aiming to embed digital tools more consistently and effectively into pharmacy practice.
15 Apr 2026
Read appraisal →International journal of medical informatics
Clinician preferences for explainable AI in critical care: a comparative study of interpretable models and visualizations for intubation decision support
BACKGROUND: The complexity of many AI models hinders their clinical adoption because the clinicians using them do not regard them as transparent. This study addresses the lack of clinician-centered explainable AI (XAI) interfaces by designing and evaluating intuitive visual explanations for intubation prediction, testing the hypothesis that workflow-compatible designs enhance acceptance. OBJECTIVE: This study compares three, time-aware, visual explanations for XAI-based intubation prediction and evaluate their acceptance, comprehension, and perceived utility among clinicians. METHODS: We developed machine learning models to estimate the near-term risk of deterioration in the patient's condition which may lead to mechanical intubation using ICU time-series data. We generated global and local explanations using SHAP and designed three customized visual formats-a temporal force plot, a temporal bar chart, and a dual-encoded SHAP heatmap. Clinicians (n = 206) evaluated comprehension and usability using objective questions and a Likert-based survey. RESULTS: Based on 4608 critically ill patients with 10 medical variables over 7 hours of data for each patient, the Random Forest (RF) model achieved the highest area under the curve (AUC): 0.94. Furthermore, the local explanations were customized and evaluated by 206 clinicians through a survey conducted on the Prolific platform. A customized heatmap representation was selected as the visualization with the highest perceived clinical utility and alignment with clinical workflows. DISCUSSION: The reported findings support the need for explanation formats to be tailored to clinical reasoning and task context, supporting the concept of cognitive fit. The heatmap's close alignment with clinicians' mental models and its graphical integrity enhances interpretability and trust. This study demonstrates that explanation effectiveness depends on contextual relevance, rather than a universal standard, and that the presentation format itself significantly shapes clinicians' trust in XAI systems. CONCLUSION: This study advances clinical XAI by introducing a time-aware explanation framework for ICU intubation decisions. By integrating temporal trends with model reasoning, our visualizations closely align with clinicians' cognitive workflows. Rigorous clinician-centered evaluation identified the dual-encoded SHAP heatmap as the most useful and workflow-compatible visualization, highlighting the importance of explanation design alongside predictive accuracy for clinical adoption.
15 Apr 2026
Read appraisal →BMJ health & care informatics
Supporting street medicine through medical informatics: a manifesto on research needs
Homeless individuals face major barriers in accessing regular healthcare, leading to the development of street medicine as a distinct humanitarian field. To enhance continuity and quality of care, consistent documentation is crucial. However, electronic health records are rarely used in European street medicine. An interdisciplinary European workshop identified four key research fields for medical informatics in street medicine: (1) Research ethics: The creation of ethical guidelines that promote co-creation and active communication with patients. (2) System development: Designing a secure, anonymous and European Health Data Space-compliant personal Electronic Health and Social Record (pEHSR), establishing standardised health and social care datasets, and implementing dynamic consent management methods. (3) Education and training: Developing targeted programmes to improve digital and health literacy among street medicine clients and training professionals in the effective use of the pEHSR system. (4) Management and evaluation: Creating management structures and evaluation frameworks suited to the unique challenges of street medicine.This manifesto calls for internationally coordinated scientific efforts to build effective, integrated digital health solutions for vulnerable populations. It encourages researchers and practitioners from medical informatics and related fields to engage with these priorities and support innovation that advances equitable healthcare access.
15 Apr 2026
Read appraisal →PloS one
Use of administrative data for evaluating trends in medically-attended Lyme disease, Manitoba, Canada, 2010-2021
While Lyme disease (LD) is the most common tick-borne disease reported to public health surveillance in North America and is increasingly recognized as a public health threat in Canada, it's incidence may be underreported. Our study aims to estimate medically-attended LD incidence in Manitoba, Canada, using administrative health data. We identified medically-attended LD cases in Manitoba from 2010 - 2021 in a claims database (Manitoba Population Research Data Repository), which contains the health records of >95% of the residents of Manitoba, using diagnostic codes, antibiotic dispensations, and laboratory results. The incidence of claims-based LD cases ranged from 8.4 to 28.5 per 100,000 population per year, 5.1 to 11.0 times higher than the incidence of surveillance-reported LD cases. The incidence of claims-based LD cases was particularly higher than the incidence of surveillance-reported LD cases in females and health regions with a low surveillance-reported incidence. Our study suggests that medically attended LD is more common than reported in surveillance. Further study is required to identify barriers to reporting. Interventions are needed to reduce the substantial burden of LD in Manitoba.
15 Apr 2026
Read appraisal →JMIR formative research
Comparing Large Language Models and Traditional Machine Translation Tools for Translating Medical Consultation Summaries: Quantitative Pilot Feasibility Study
BACKGROUND: Translation of medical consultation summaries is essential for equitable health care communication in culturally and linguistically diverse populations. While machine translation (MT) tools and large language models (LLMs) are widely accessible, their feasibility and safety for health care contexts remain underexplored. OBJECTIVE: This pilot study investigates the feasibility and limitations of using LLMs and traditional MT tools to translate medical consultation summaries from English into the most common languages other than English spoken in Australia-Arabic, Chinese (simplified written form), and Vietnamese. METHODS: Two simulated summaries-a simple patient-facing summary and a complex clinician-oriented interprofessional letter-were translated using 3 LLMs (GPT-4o, Llama-3.1, and Gemma-2) and 3 MT tools (Google Translate, Microsoft Bing Translator, and DeepL). Translations were benchmarked against professional third-party interpreter translations using Bilingual Evaluation Understudy, Character-level F-score, and Metric for Evaluation of Translation with Explicit Ordering metrics. RESULTS: The translation performance varied across languages, tools, and summary complexity when assessed using automatic evaluation metrics. Traditional MT tools outperformed LLMs on surface-level metrics, while LLMs showed relative strengths in semantic similarity for Vietnamese and Chinese. Arabic translations improved with complex input, suggesting morphological advantages. The metric-based evaluation highlighted feasibility but also risks, particularly in Chinese clinical contexts. CONCLUSIONS: This pilot study provides formative evidence of opportunities and limitations in applying artificial intelligence translation for health care communication. Findings underscore the importance of human oversight; domain-specific evaluation metrics; and further formative and clinical research to guide the safe, equitable use of artificial intelligence translation tools.
14 Apr 2026
Read appraisal →JMIR aging
Personalized Digital Care Program Allocation for Older Adults: Reinforcement Learning-Based Simulation Study
BACKGROUND: As the demand for innovative older adult care grows alongside a shortage of care workers, personalization is key to optimizing services and enhancing long-term sustainability. This study proposes an adaptive reinforcement learning (RL)-based framework to promote precision digital care by dynamically assigning care programs based on individuals' unique characteristics and evolving needs. Its effectiveness was evaluated through simulation-based experiments comparing multiple allocation methods within an artificial intelligence (AI)-powered care call service for older adults. OBJECTIVE: This study aimed to develop and evaluate an RL-based model for personalizing digital care program allocation to optimize care engagement and health outcomes among low-income older adults living alone. METHODS: We developed the framework by using contextual bandits, specifically Thompson Sampling, to maximize user outcomes. Four program allocation strategies were tested using a synthetic dataset of user features and program attributes. Simulations were conducted over multiple iterations to evaluate how the model adapts over time and optimizes program assignments compared with static methods. RESULTS: Four program allocation methods were compared across 100 simulation runs (n=3000 assignments per run) using 2 datasets (AI Call: n=1196; Community Health Survey [CHS]: n=72,812): (1) systematic allocation (baseline), (2) single best program based on population average, (3) idealized personalized delivery (theoretical upper bound), and (4) precision digital care using Thompson Sampling. Precision digital care outperformed baseline and population-average approaches, achieving outcomes comparable to the theoretical upper bound. Compared to systematic allocation, call success rates increased by 84.2% (AI Call) and 54.4% (CHS), Patient Health Questionnaire-2 depression scores decreased by 32.1% (AI Call) and 41.4% (CHS), and self-reported health scores improved by 19% (AI Call) and 22% (CHS). It also showed improved learning efficiency, refining assignments dynamically as it learned from user responses. CONCLUSIONS: Our findings emphasize the importance of personalization in digital care. We plan to refine and validate the model through a publicly funded AI care program for community-dwelling, low-income older adults living alone in the Republic of Korea. RL offers a scalable and effective approach to advance precision digital care delivery and support future innovations in aging services.
14 Apr 2026
Read appraisal →JMIR cancer
Leveraging Innovative Electronic Health Record Data to Characterize Social Determinants of Health Among Survivors of Cancer in Persistent Poverty Areas: Cross-Sectional Study
BACKGROUND: Residents in persistent poverty areas experience higher cancer mortality due to social determinants of health that negatively affect multiple factors, including health behaviors. OBJECTIVE: This study aimed to characterize demographic, clinical, and social determinant of health factors among survivors of cancer in persistent poverty areas using electronic health record (EHR) data-including an embedded social risk screener and natural language processing (NLP) of social work notes-to inform community-engaged adaptation of lifestyle interventions. METHODS: EHR data from a large multispecialty group practice were extracted for patients with cancer residing in zip codes inclusive of persistent poverty areas targeted for a health behavior intervention and receiving care between January 2018 and November 2023. Self-reported social determinant of health data were obtained using the Protocol for Responding to and Assessing Patients' Assets, Risks, and Experiences (PRAPARE) and through NLP of social histories from a social work visit. RESULTS: We identified 2672 unique patients with cancer, of whom 578 (21.6%) had PRAPARE data and 1597 (59.8%) had social history data available for analysis. The most common cancers among survivors (n=1420, 53.1% female; n=1299, 48.6% Black individuals; mean age 65.2, SD 13.7 years) included breast (n=536, 20.1%), prostate (n=400, 15%), and lymphoid or hematopoietic (n=323, 12.1%) cancer. Among survivors in persistent poverty areas (n=509, 19%; all with a high Social Vulnerability Index), 34.6% (176/509) were single, 55.4% (282/509) had Medicare coverage (with only 73/509, 14.3% having private insurance), 36.5% (186/509) had obesity, 63.9% (325/509) had hypertension, and 31.2% (159/509) had diabetes. Of survivors in persistent poverty areas with PRAPARE data, 15.8% (19/120) lacked transportation, 4.2% (5/120) lived with housing insecurity, and 6.7% (8/120) felt unsafe where they lived. CONCLUSIONS: Innovative EHR and NLP approaches identified several socioeconomic and safety-related challenges along with opportunities for health behavior interventions to leverage Medicare coverage and target multiple comorbidities when adapting interventions for survivors of cancer living in persistent poverty areas.
14 Apr 2026
Read appraisal →JMIR medical informatics
Responsible AI for Predicting Delayed Hospital Discharge Among Older Adults: Development and Evaluation Study for Balancing Accuracy, Equity, and Explainability
BACKGROUND: Amid growing demands and constrained health care resources, effective hospital bed capacity management is crucial. Delayed hospital discharge, where patients remain in the hospital beyond the need for acute care, strains resources, affects patient outcomes, and reduces system efficiency. Predicting such delays facilitates early interventions to avert them and alleviate burdens on patients, care partners, hospitals, and the broader health care system. OBJECTIVE: This study aimed to develop comprehensive predictive analytics for delayed discharges among older adults using explainable machine learning to boost transparency and interpretability, while integrating fairness to mitigate algorithmic biases. METHODS: Leveraging longitudinal data from over 2 decades in Ontario, Canada, we applied extreme gradient boosting and logistic regression models to predict delayed discharges within 90 days post-acute care. Data preprocessing included a 2-year look-back for clinical histories and balanced sampling to address class imbalance. Model performance was assessed via area under the receiver operating characteristic curve, calibration, and clinical utility. Fairness was evaluated across sex, urban or rural residence, and residential instability using several threshold-free metrics. Explainability was examined at the global model level (via partial dependence plots and permutation feature importance) and locally (via Shapley Additive Explanations, breakdown, and ceteris paribus methods), with principal component analysis used to cluster key features for high-risk patients. RESULTS: The extreme gradient boosting model outperformed logistic regression, achieving an area under the receiver operating characteristic curve of 0.82 on the test set, with acceptable within-group and cross-group ranking fairness across subgroups. Explainability clustering analyses identified functional and cognitive declines (eg, care support needs, dementia, and mobility issues) and regional disparities as primary drivers of high-risk predictions. Bias mitigation improved calibration parity, especially when stratifying by residential instability, underscoring the trade-offs policymakers must weigh between accuracy, fairness, and explainability. CONCLUSIONS: This study demonstrates the potential of responsible artificial intelligence in health care, emphasizing the need to balance predictive accuracy, equity, and interpretability. It uncovers systemic gaps and offers actionable insights for enhanced discharge planning, resource optimization, and equitable care delivery.
14 Apr 2026
Read appraisal →Medicine
Bibliometric and visual analysis of artificial intelligence-related research in osteoarthritis: Trends and frontiers (2008-2025)
Osteoarthritis (OA) causes joint pain, functional impairment, and significantly impacts patients' quality of life. Recent advancements in artificial intelligence (AI) technology have provided new opportunities for OA diagnosis, treatment, and management. This study aims to understand the knowledge structure and research frontiers of "AI and OA" using bibliometric methods. Literature on "AI and OA" from January 1, 2008, to March 31, 2025, was retrieved from the Web of Science Core Collection database. VOSviewer, CiteSpace, and the R package "bibliometrix" were used for analysis. The study included 659 articles from 59 countries. Since 2019, the number of scientific outputs related to OA and AI has increased annually. The United States, China, and the United Kingdom are the main contributing countries, with Harvard University, Seoul National University, and the University of California system as leading institutions. The "JOURNAL OF ARTHROPLASTY" and "OSTEOARTHRITIS AND CARTILAGE" are the most contributing journals. Among the 2404 authors, GUERMAZI A and ROEMER FW are the most influential. Thematic trend analysis shows that AI and machine learning are driving new trends in OA research, especially in diagnosis, treatment planning, and patient monitoring. This paper summarizes the bibliometric trends and development in "AI and OA" studies over 18 years, identifying the latest research frontiers and hot topics. It provides valuable references for related research and highlights the importance of interdisciplinary collaboration as research topics diversify.
12 Apr 2026
Read appraisal →PloS one
Privacy-preserving multimodal federated learning pipeline for cyber-resilient healthcare systems
The integration of Internet of Things (IoT) devices and electronic medical records (EMRs) has transformed healthcare delivery but has also created new vulnerabilities to cyberattacks that threaten both data confidentiality and patient safety. Conventional centralized machine learning approaches for intrusion detection are impractical in this domain due to strict privacy regulations, heterogeneous data sources, and the risk of single points of failure. To address these challenges, we propose a secure distributed machine learning pipeline for cyber-resilient healthcare systems. The framework combines federated optimization with split learning for sensitive EMR data, robust aggregation to mitigate poisoned updates, and differential privacy with secure aggregation to protect against inference attacks. Multimodal fusion is enabled through temporal consistency regularization for IoT traffic and cross-layer contrastive alignment to link EMR representations, ensuring improved anomaly detection across diverse healthcare environments. Experiments conducted on representative IoT and EMR datasets demonstrate that the proposed pipeline achieves accuracy of 0.942 on IoT data, 0.931 on EMR data, and 0.953 in the combined setting, with corresponding F1-scores of 0.921, 0.908, and 0.932. Ranking metrics further confirm superiority with AUROC up to 0.961 and AUPRC up to 0.947, outperforming deep baselines by margins of +0.025 to +0.033. Robustness analysis shows graceful degradation under client poisoning ([Formula: see text] at 30% malicious clients) and resilience under severe communication constraints (accuracy [Formula: see text] at 90% update sparsification). Detection latency is reduced to an average of 5.9 time steps, compared to 7.8 for the strongest deep baseline. These results highlight that secure distributed pipelines can deliver both strong detection capabilities and regulatory compliance, providing a practical path toward safeguarding next-generation healthcare infrastructures against evolving cyber threats.
12 Apr 2026
Read appraisal →Journal of medical Internet research
Large Language Model-Based Analysis of Statin Therapy Discussions and Sentiment on Social Media: Cross-Sectional Observational Study
BACKGROUND: Statin therapy, despite proven cardiovascular benefits, remains underused. Social media platforms may capture patient perspectives that are less visible in clinical encounters. OBJECTIVE: This study aimed to characterize themes, sentiment, and decision-making factors related to statin therapy through large language model (LLM)-based analysis of Reddit discussions. METHODS: This cross-sectional observational study analyzed English-language Reddit posts and comments mentioning statins from January 2022 to May 2025, identified via keyword-based Reddit application programming interface searches (≤1000 posts per keyword). A total of 5328 retrieved discussions (n=1661, 31.2% posts and n=3667, 68.8% keyword-containing comments) from public subreddits were included. Themes, sentiments (positive, neutral, or negative), guideline-informed clinical relevance, information-seeking behavior, adverse effect mentions, decision factors, and adherence-related content were extracted using an LLM-based pipeline. RESULTS: Among 5328 discussions, prominent topics included adverse effects (n=1697, 31.9%), decision-making references related to laboratory results and physician advice (n=2767, 51.9% and n=2034, 38.2%, respectively), and alternative approaches (n=2485, 46.6%). Overall sentiment was neutral in 34% (n=1812) of discussions, negative in 30.9% (n=1646), and positive in 16.9% (n=900); the remainder were mixed or unclear. Statin-directed sentiment was neutral in 44.1% (n=2350) of discussions, negative in 25.2% (n=1343), and positive in 12.5% (n=666); the remainder did not express statin-directed sentiment. High clinical relevance was identified in 12.6% (n=672) of discussions. Adherence-related issues were mentioned in 29.8% (n=1587) of discussions. Among adverse effect mentions, muscle pain (n=129, 7.6%) and fatigue (n=110, 6.5%) were common. CONCLUSIONS: LLM-enabled analysis of Reddit discourse highlights substantial negative sentiment, adherence-related concerns, and adverse effect narratives surrounding statin therapy. These findings suggest opportunities for patient-centered communication and shared decision-making strategies that address symptom attribution, uncertainty, and information needs in digital information environments.
12 Apr 2026
Read appraisal →Journal of medical Internet research
Beyond GPT-4: The Rapidly Evolving Potential of Large Language Models for Clinical Guideline Improvement
This commentary reviews the study by Jones et al, which evaluated whether GPT-4 could improve the readability of injectable medication guidelines while preserving important safety information. The study found that GPT-4 produced modest readability gains comparable to manual revision, but also introduced omissions and meaning changes in a minority of sections. These findings highlight both the potential and limitations of early large language models (LLMs) in clinical contexts. However, this study reflects the capabilities of a specific model in a rapidly evolving domain. Since the release of GPT-4, advances in multistep reasoning, model-critique workflows, and structured validation have substantially improved the ability of newer systems to detect omissions, maintain factual fidelity, and support controlled editing. As a result, some documented limitations may stem from the constraints of a single-model, single-pass workflow rather than intrinsic flaws in LLM-assisted guideline revision. This commentary highlights the need for evaluation frameworks that can keep pace with LLM progress and emphasizes that clinical oversight and user-centered testing remain essential. Updated research using contemporary models is needed to determine how emerging architectures can more safely support clarity, consistency, and maintenance of clinical guidelines.
12 Apr 2026
Read appraisal →PloS one
Applying machine learning and natural language processing to patient safety event reports: Identifying patterns of cardiovascular diagnostic errors
OBJECTIVE: To explore the utility of natural language processing (NLP) and machine learning (ML) techniques to identify unsafe conditions leading to cardiovascular diagnostic errors using patient safety event (PSE) reports data. METHODS: PSE reports from January 2016 to August 2021 from a multi-hospital healthcare system in the mid-Atlantic region of the United States were included in this study. To have the true cardiovascular diagnostic errors labels for PSE reports, each individual PSE report was manually reviewed to find clinician-reported narratives describing current definitions of cardiovascular diagnostic errors. The PSE reports which contained cardiovascular diagnostic errors-related narratives were labeled as one and zero otherwise. Four binary ML models were employed to identify cardiovascular diagnostic errors narratives and common features from annotated PSE reports data: (1) simple logistic regression, (2) elastic net, (3) XGBoost, and (4) deep neural networks. RESULTS: XGBoost outperformed the rest of the models in identifying cardiovascular diagnostic errors -related reports and achieved high performance metrics on the testing data (AUROC = 0.914, specificity = 0.982, PPV = 0.866, accuracy = 0.929, F-1 score = 0.738, and AUPRC = 0.783). Pacemaker emerged as a significant signal for cardiovascular diagnostic errors in our PSE reports. Our analysis demonstrated that ordering MRI for patients with pacemaker was a frequent theme among cardiovascular diagnostic errors-related PSE reports containing the word pacemaker. Order, EKG, cardiac, and chest were the five most important features in identifying cardiovascular diagnostic errors events. These words were utilized in explaining heart conditions, associated care process, and related safety incidents. CONCLUSIONS: Findings from our study demonstrates the feasibility of ML and NLP techniques in identifying cardiovascular diagnostic errors-related reports among the existing PSE data sources. However, validation in external healthcare systems is needed before broader application.
11 Apr 2026
Read appraisal →Journal of medical Internet research
Views of People With Psychosis About Algorithm-Based Relapse Prediction and Data Sharing: Qualitative Study
BACKGROUND: Preventing relapses of psychosis is difficult and important. Digital remote monitoring (DRM) systems are being developed and tested to support this. Increasingly, these systems use algorithm-based relapse prediction. Hence, understanding stakeholder views about algorithmic prediction is crucial. Existing qualitative work has explored health professionals' views, but very few studies have examined the perspectives of people with psychosis on this topic. OBJECTIVE: This paper aimed to provide an in-depth examination of the views of people with psychosis regarding algorithmic relapse prediction within a DRM system that incorporates active symptom monitoring and passive sensing data. METHODS: People with psychosis (n=58) were recruited from 6 geographically distinct areas of the United Kingdom. They participated in semistructured qualitative interviews exploring their views about using a DRM system that predicts psychosis relapse based on a machine learning algorithm. Transcripts were analyzed using reflexive thematic analysis. People with lived experience of psychosis were involved extensively in study design, analysis, and reporting. RESULTS: Findings were described across 4 themes. First, accuracy was a prominent theme. Participants emphasized that transparency about algorithm sensitivity and specificity is crucial and discussed the risks of the relapse prediction algorithm producing false positives (flagging that someone was relapsing when they were not) and false negatives (missing actual relapses). In both cases, participants said that errors may be partially mitigated through a human-in-the-loop approach (theme 2), with DRM blended with human oversight, from clinicians or a dedicated digital monitoring team, and calibrated based on service user, carer, and clinician feedback. The third theme, trust, fears, and choice, noted the interplay between users' trust in the DRM system and their relationship with the clinical team. This theme described participants' fears about potential overreactions (hospitalization or excessive medication) or underreactions (no additional support) from the clinical team in response to algorithm-generated relapse predictions. It emphasized the importance of retaining choice around the use of relapse detection algorithms and the sharing of personal data. The final theme described participants' views about the benefits of using a relapse prediction algorithm, including facilitating early intervention, triaging care according to need, minimizing human bias in assessment, and efficiency in saving staff time. CONCLUSIONS: People with psychosis acknowledged potential benefits of algorithm-assisted relapse prediction for receiving timely or efficient care, but with several caveats. Algorithm-generated relapse alerts need to be sufficiently accurate and must be interpreted, with understanding of their limitations, by a trustworthy human who is aware of the relevant context. Algorithm-based relapse predictions should only be used with valid consent, in a way that promotes and respects the autonomy and voice of service users and avoids increasing the use of excessive restriction.
11 Apr 2026
Read appraisal →JMIR aging
Facilitators and Barriers to Digital Self-Management in Older Adults With Depression: COM-B and Theoretical Domain Framework Qualitative Study
BACKGROUND: Depression in older adults presents unique challenges in self-management. Digital tools, such as mobile health (mHealth) apps, have the potential to support this population. This study explored the facilitators and barriers to digital self-management in older adults with depression to inform the design of effective mHealth apps. OBJECTIVE: This study aims to explore the facilitators and barriers to digital self-management in older patients with depression with the aim of informing the design and development of mHealth apps for older adults. METHODS: A purposive sampling method was used to recruit 25 older patients with depression from July to September 2024. Semistructured interviews were conducted to capture real-life experiences. Directed content analysis ensured objective and accurate data interpretation, and the Capability, Opportunity, Motivation, and Behavior (COM-B) model and the Theoretical Domains Framework were applied to identify facilitators and barriers related to behavior. RESULTS: Six themes were identified based on the COM-B model and the Theoretical Domains Framework: perception of illness and accumulation of personal experience; dual challenges of cognitive abilities and physical limitations; integration of digital technologies and acceptance differences; social influences and access to and utilization of support resources; environmental constraints and accommodations; and intertwined influences of beliefs, emotions, and motivation. A total of 13 barriers and 11 facilitators were identified. CONCLUSIONS: Digital self-management in older adults with depression is complex and influenced by multiple interrelated factors. Effective mHealth apps must integrate the cognitive, emotional, and social contexts of patients to provide user-friendly, personalized solutions.
11 Apr 2026
Read appraisal →Journal of medical Internet research
Digital Health Technology Use Among Rehabilitation Professionals in China: Multi-Province Cross-Sectional Survey
BACKGROUND: The rapid expansion of rehabilitation needs in China has intensified pressure on a workforce that remains unevenly distributed. Digital health technologies (DHTs) offer potential to increase service reach and efficiency. However, little is known about how rehabilitation professionals currently gather and document clinical information, nor about their readiness to integrate digital tools into routine practice within China's rapidly digitalizing health system. OBJECTIVE: This study aimed to describe how rehabilitation professionals in China collect subjective and objective clinical information, document patient data in routine practice, and assess their willingness to use DHTs in clinical settings. METHODS: We conducted a multi-province observational cross-sectional survey using a culturally adapted questionnaire based on the World Health Organization Digital Health Interventions framework. The instrument assessed participant characteristics, information collection methods, documentation practices, and willingness to adopt digital functions across rehabilitation activities. Descriptive analyses and subgroup comparisons were performed on 324 complete responses from certified rehabilitation professionals. The multi-province cross-sectional online survey was conducted among licensed rehabilitation professionals in China with internet access. Participants were recruited through professional networks and social media platforms. RESULTS: Respondents were drawn from 20 provincial-level administrative regions across China, including Fujian (n=72), Guangdong (n=77), and Shanxi (n=45), among others, with 82.7% (268/324) employed in public sector rehabilitation services. Traditional methods dominated clinical work. Face-to-face communication was used frequently for subjective assessment by 96.3% (312/324) of respondents, whereas digital channels such as email (22/324, 6.8%) and telephone (47/324, 14.5%) saw limited use. For objective information, visual observation (271/324, 83.7%) and manual measurement tools (195/324, 60.2%) remained the primary approaches, while motion capture technology (45/324, 13.8%) and wearable sensors (13/324, 4%) were rarely used. Documentation practices also relied heavily on analogue formats, with 82.1% (266/324) using handwritten notes and 60.2% (195/324) using paper templates. In contrast, willingness to adopt DHTs was consistently high, with 80.6% (261/324) of respondents indicating readiness to use digital systems for identity verification, 79.0% (256/324) for progress tracking, and 78.1% (253/324) for outcome measurement. Subgroup analyses revealed that educational level significantly influenced the adoption of advanced technologies, with master's or doctoral degree holders reporting higher use of sensor-based assessment, motion capture, and wearable devices. In contrast, professional title and clinical specialty showed limited influence, with no significant differences observed for most digital health functions. CONCLUSIONS: Rehabilitation professionals in China demonstrate strong readiness to use DHTs, yet their routine practice remains largely paper-based and analogue. These findings provide evidence to inform implementation strategies, workforce training, and system-level planning aimed at accelerating digital transformation in rehabilitation services.
11 Apr 2026
Read appraisal →Acta dermato-venereologica
Predicting Melanoma Impact on the Swedish Healthcare System from the Adult Population Using Machine Learning on Registry Data
Melanoma incidence has increased in Western countries over the past 50 years, leading to significant healthcare costs. In Sweden, comprehensive healthcare registries enable large-scale prediction studies using machine learning. Several machine learning models were evaluated to predict melanoma diagnoses using Swedish registry data, assessing the added value of diagnostic and medication data beyond demographics. The study included all adults in Sweden with continuous residency for 9.5 years (n = 6,036,186). The outcome was a melanoma diagnosis, including melanoma in situ, recorded within 5 years after the index date (31 December 2014). Predictors included age, sex, income, education, marital status, region of birth, diagnoses, and dispensed drugs. Models tested were logistic regression, gradient boosting, random forests, and a neural network. A total of 38,582 individuals (0.64%) developed melanoma. The gradient boosting model using all predictors performed best, with an area under the receiving operating characteristics curve (AUC) of 0.735 (95% confidence interval [CI], 0.725-0.746). When diagnosis and medication data were excluded, AUC dropped to 0.681 (95% CI: 0.670-0.691). The findings highlight that including healthcare codes improves predictive performance, and demonstrate the utility of Swedish registries for computational phenotyping. This approach may support early detection of melanoma and targeted follow-up.
10 Apr 2026
Read appraisal →JMIR cancer
Empathic AI for Patient-Centered Cancer Care: A Scoping Review of Patient Navigation, Support, and Clinical Practice
BACKGROUND: Artificial intelligence (AI) is rapidly reshaping oncology, offering advancements in clinical care and patient support. A growing area of interest is the integration of empathic AI: systems integrating clinical precision with emotional intelligence to support medical decision-making and the emotional and psychosocial well-being of patients and caregivers. OBJECTIVE: This review aimed to explore the role of empathic AI in cancer care, with a focus on its applications in patient education, clinician support, and emotional care. It also evaluated the ethical, cultural, and implementation challenges associated with its integration into clinical practice in oncology. METHODS: A systematic search of literature was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines. Peer-reviewed papers published between January 2018 and January 2025 were identified through a search of PubMed, Scopus, and IEEE Xplore, and citation tracking. Eligible studies focused on applications of empathic AI in oncology. A total of 44 studies were included and analyzed thematically across 3 core clinical domains: tailored communication and education, diagnostics and care plan optimization, and emotional and psychosocial support. RESULTS: Empathic AI demonstrates the potential to improve cancer care by enhancing patient education, clinical decision-making, and emotional support. Common applications include personalized education for patients and providers, support for diagnostic and therapeutic decisions, and tools designed to recognize and respond to patient distress. Several studies noted improved patient engagement and reduced clinician burden. However, concerns were identified regarding overreliance on AI systems, cultural insensitivity, and patient privacy. CONCLUSIONS: Empathic AI represents a promising advancement in patient-centered oncology, integrating emotional intelligence into clinical care. Its successful implementation will require careful attention to ethical, cultural, and clinical considerations to ensure health equity, trust, and safety in AI-assisted cancer care.
10 Apr 2026
Read appraisal →Journal of medical Internet research
Innovations in Deaf Health Care Communication: Systematic Review of Sign Language Recognition Systems
BACKGROUND: Deaf individuals often face communication challenges when interacting with those who can hear. Within health care settings, these challenges may pose risks to their safety, potentially resulting in misdiagnoses, treatment errors, and decreased quality of care. OBJECTIVE: This study aims to systematically review the evidence on communication systems reported in the literature that use human-computer interaction techniques to support communication between deaf individuals who use sign language and hearing health professionals in health care settings. The review focuses on systems that are either currently in use or proposed for use in health care and that have been tested using human participants or videos of human users. METHODS: A comprehensive search was performed via MEDLINE, Web of Science, ACM, IEEE Xplore, Scopus, and Google Scholar in March 2025. The inclusion criteria comprised studies developing a sign language recognition system within a health care context and testing with human users. Eligible studies underwent screening by 2 independent investigators (LRV and LMMSR or LFRdO and GTdSS), with any disagreements resolved by a senior researcher (MSM). RESULTS: The search retrieved 21,778 publications, and screening of reference lists identified 2 additional studies, resulting in a total of 23 studies meeting the eligibility criteria. Most systems (15/23, 65.2%) were image-based, while 34.8% (8/23) relied on sensors (glove-based or depth-sensing). Applications varied across health care settings, including general hospital care (10/23, 43.5%), emergencies (8/23, 34.8%), and primary care (4/23, 17.4%). All systems were in the development and testing stage, with no data on security and psychological impacts. Accuracy ranged from 25% to 100% for image-based and 72% to 99.7% for sensor-based systems. Bidirectionality and facial expression recognition, crucial for effective communication, were largely overlooked. CONCLUSIONS: Image-based systems were more common than sensor-based ones, though both showed wide variability in accuracy in recognizing and interpreting signs. Most systems failed to address critical aspects such as bidirectional communication and the recognition of facial expressions, essential for effective communication. None fully addresses the requirements for integration into health care settings. These findings highlight the need for further research on implementation, usability, and impact on the quality of care for deaf patients.
10 Apr 2026
Read appraisal →JMIR mHealth and uHealth
Trust and Use of Recommendations for Health Apps Among European Residents: Cross-Sectional Survey
BACKGROUND: There is growing recognition of the role of health apps in addressing health care system challenges, yet app quality and evidence vary widely, and consumers have little decision support at the point of download. OBJECTIVE: This study aimed to explore which sources of recommendations European residents use and trust when choosing health apps and whether residents support government review and rating of health apps. METHODS: We conducted a cross-sectional online survey (December 7, 2022, to February 16, 2023) in 26 languages targeting residents of the European Economic Area, the United Kingdom, and Ukraine. The survey contained 11 questions covering demographics, types of apps used, sources of advice used and trusted, and views on government review and rating. We included only fully completed responses (N=1228). Descriptive statistics are presented as counts and percentages. For subgroup analyses, we dichotomized trust responses ("I trust" vs other responses) and tested associations with gender (Fisher exact test), age group, and education level (chi-square test); an adjusted significance threshold was applied (P<.004) to account for multiple testing. We followed the CHERRIES (Checklist for Reporting Results of Internet E-Surveys) guidelines for reporting online surveys and STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) recommendations for observational studies. RESULTS: A total of 1228 respondents from 33 countries completed the survey; 1110 (90.4%) reported using one or more health apps. COVID-19 apps (763/1228, 62.1%) and activity apps (737/1228, 60.0%) were the most frequently used, whereas disease management (86/1228, 7.0%), diagnostic (77/1228, 6.3%), and treatment apps (70/1228, 5.7%) were the least used. Sources used to choose apps included family and friends (434/1228, 35.3%), health professionals (412/1228, 33.6%), and government or health authorities (358/1228, 29.2%). The most trusted sources were health professionals (987/1228, 80.4%), pharmacists (750/1228, 61.1%), and government or health authorities (736/1228, 59.9%). No statistically significant differences in trust by gender were observed (all P>.0038). Some differences by age and education were observed for select sources (eg, government or health authorities: χ21=8.546; P=.003; family and friends: χ2=19.133; P<.001). Overall, 1060 (86.3%) of 1228 respondents supported government review and rating of health apps (either directly or by commissioning another organization). CONCLUSIONS: In this large multilingual European survey, most respondents reported experience with health apps and placed greatest trust in health professionals, pharmacists, and government or health authorities, yet professional recommendations were used less often than informal sources. There is clear public support for government-led review and rating schemes to guide consumer choice. Efforts to make trustworthy, easy-to-find information available at the point of download and to support health care professionals in recommending high-quality apps could help bridge the gap between trusted and used sources.
10 Apr 2026
Read appraisal →Journal of medical Internet research
The Alberta Quality Assessment Tool: Risk of Bias (AQAT:RoB) for the Evaluation of Medical Large Language Model Question-Answer Studies: Development and Pilot Validation
BACKGROUND: Despite the transformative potential of large language models (LLMs) in health care, the rapid development of these tools has outpaced their rigorous evaluation. While artificial intelligence-specific reporting guidelines have been developed to address standardized reporting of artificial intelligence studies, there is currently no specific tool available for risk of bias assessment of LLM question-answer (QA) studies. Existing risk-of-bias tools for medical research are not well suited to the unique challenges of evaluating LLM-QA studies, which creates a critical gap in assessing their safety and effectiveness. OBJECTIVE: This study aims to develop the Alberta Quality Assessment Tool: Risk of Bias (AQAT:RoB) for LLM-QA studies to systematically evaluate the validity and risk of bias in LLM-QA studies. METHODS: We conducted 2 literature reviews. The first was on quality assessment tools for LLM-QA studies, and the second was on LLM-QA studies, which informed the first draft of the AQAT:RoB. The draft AQAT:ROB was further refined through a prespecified iterative process of modified Delphi, consensus meeting, and validation. The first Delphi process occurred between May 1 and May 20, 2025, and the first consensus meeting was held on May 22. The first round of validation was completed by 4 evaluators, who were not part of the consensus meeting, on 16 randomly selected studies. As this first round of validation surpassed our a priori threshold of ≥80% agreement and a Cohen κ of ≥0.61 between evaluators, no further rounds of development and validation were undertaken. A second Delphi process occurred between February 20 and February 23, 2026, to vote on postpilot changes in response to peer review. RESULTS: The AQAT:RoB consists of 5 high-level domains (Questions, Reference Answers, LLM Answers, Evaluators, Outcomes). These domains are subdivided into 9 subdomains. Each subdomain includes at least one "Support for Judgment" and at least one "Type of Bias" and is to be rated "low," "high," or "unclear" for risk of bias. A pilot evaluation was completed by internal validators who were not part of the consensus discussion and were asked to complete the AQAT:RoB form for each assigned study. Each of the 16 studies was evaluated by 2 evaluators independently. Pilot validation showed a percent agreement of 86.1% and a Cohen κ of 0.70 between assessors. CONCLUSIONS: The AQAT:RoB demonstrates promising initial reliability for assessing the validity or risk of bias in LLM-QA studies. The tool will benefit from future refinements, external validation, and periodic updates to keep pace with evolving technology.
10 Apr 2026
Read appraisal →PloS one
Enhancing intrusion detection in encrypted DoH traffic through a robust ensemble learning framework
The DNS over HTTPS (DoH) protocol enhances user privacy by encrypting DNS queries and responses within HTTPS. However, this encryption enables attackers to tunnel malicious traffic through covert DoH channels and make detection difficult for network security and enterprises. In this study, we propose a stacked ensemble approach that employs four sequential base learners' - two Long Short-Term Memory (LSTM) and two Gated Recurrent Unit (GRU) neural networks as base classifier networks with varied hyperparameters and XGBoost (eXtreme Gradient Boosting) as the meta-learner. An XGBoost model trained on out-of-fold predictions generated by base learners supported through stratified 5-fold cross validation. We also applied Recursive Feature Elimination (RFE) with XGBoost for cross-validated feature selection and decreased high-dimensional feature spaces from 29 to 13 and 20 features on the CIRA-CIC-DoHBrw-2020 and DoH-Tunnel-Traffic-HKD datasets respectively. The proposed framework was evaluated on the CIRA-CIC-DoHBrw-2020 and DoH-Tunnel-Traffic-HKD datasets achieved 0.9996 recall, 0.9999 F1-score and 1.0000 AUC-PR on CIC-DoH dataset and 0.9998 recall, 0.9999 F1-score, and 1.0000 for both AUC-ROC and AUC-PR on HKD-DoH dataset. The proposed model outperforms all individual base classifiers, evaluated ML models, and existing state-of-the-art approaches. The proposed XGBoost meta-model needs only 1.26 seconds for training and 0.129 seconds for inference on HKD-DoH dataset, thus making the proposed framework both highly accurate and computationally efficient for real-world intrusion detection in encrypted DoH traffic.The preprocessed data and the source code supporting the findings of this study are available at: https://github.com/soreettihussein/Stacked-ensemble-meta-learning-IDS-for-DoH-network-traffic/tree/main.
10 Apr 2026
Read appraisal →PloS one
Comparative analysis of text readability and writing styles in AI-generated vs. Human-written academic abstracts
Research article abstracts are vital in scientific publications for readers to assess a study's significance. The increasing use of AI tools, such as Kimi, ChatGPT and DeepSeek, to generate abstracts raises concerns about their readability and writing styles compared to human-written ones. The study aims to compare the differences in text readability and writing styles between human-written against AI-generated abstracts. A total of 150 abstracts of high-impact journal articles in the field of linguistics and computer science, 75 from each discipline, and another 150 AI-generated abstracts from the same corpus of articles served as the source texts for analysis. The Readability Scoring System, a computational tool, yielded readability and writing style metrics, while expert evaluation was performed to assess the quality of AI-generated academic abstracts. The quantitative data generated were analysed using SPSS 27 with non-parametric statistical methods. Key findings revealed: (1) AI-generated abstracts exhibited significantly lower readability across eight metrics, indicating greater complexity and lower readability; (2) Discipline-specific analysis showed five differing metrics in linguistics and eight in computer science; (3) Interdisciplinary comparisons revealed non-significant differences across nine readability metrics, highlighting AI's potential to mimic natural writing. However, it still faces challenges in generating lexically diverse content. These results underscored the current limitations of AI in generating readable and human-like abstracts, especially in technical fields.
10 Apr 2026
Read appraisal →Journal of medical Internet research
Preferences for Personalized Text Message Appointment Reminders Among Outpatients in a Universal Health System: Cross-Sectional Study
BACKGROUND: SMS text messaging reminders are widely used to reduce missed outpatient appointments; however, evidence remains limited regarding which types of reminder content patients prefer, particularly within East Asian universal health systems. In Taiwan, minimal financial barriers to care and unrestricted access to secondary and tertiary hospitals contribute to high outpatient visit volumes and persistent no-show rates. These contextual features underscore the need for behaviorally informed and demographically tailored reminder strategies rather than uniform messaging approaches. OBJECTIVE: This study aimed to examine patient preferences for 6 theory-guided SMS appointment reminder types and to identify the predictors of reminder preference related to demographic characteristics and health care utilization, with the goal of informing personalized reminder design for a forthcoming randomized controlled trial. METHODS: We conducted a cross-sectional online survey among adults in Taiwan with prior outpatient experience. Six SMS reminder prototypes were developed based on behavioral communication principles and validated by a multidisciplinary expert panel using item-level content validity indices. Participants selected their preferred SMS reminder type and reported sociodemographic characteristics and recent health care utilization. Bivariate associations were examined using chi-square tests and one-way ANOVA, with Benjamini-Hochberg false discovery rate correction applied to control for multiple testing. To identify independent predictors of SMS reminder preference while adjusting for potential confounding, we fitted a multinomial logistic regression model with all covariates entered simultaneously. RESULTS: A total of 1095 respondents completed the survey. General reminders and messages referencing prior missed appointments were most frequently preferred, whereas empathy-based or relationally framed messages were selected less often. In false discovery rate-adjusted univariate analyses, both age and sex were associated with SMS reminder preference. However, in the fully adjusted multinomial logistic regression model, age emerged as the only statistically significant independent predictor. Participants younger than 50 years were significantly more likely to prefer alternative reminder message types compared with the general reminder (adjusted odds ratio 1.64, 95% CI 1.18-2.28; P=.003). Sex did not retain statistical significance after multivariable adjustment. Other sociodemographic characteristics and health care utilization variables, including education level, employment status, residential region, outpatient visit frequency, and recent missed appointments history, were not independently associated with reminder preference. CONCLUSIONS: Preferences for outpatient SMS reminder content vary systematically, with age representing the most robust independent predictor. Across the sample, concise and behavior-focused reminders were preferred over empathy-oriented or relational formats. These findings support age-informed tailoring of SMS reminder content and provide content-validated SMS prototypes for use in subsequent interventional research. The results offer formative evidence to guide the design of randomized trials aimed at reducing outpatient no-shows and improving the efficiency of ambulatory care delivery in Taiwan's universal health care system.
10 Apr 2026
Read appraisal →JMIR human factors
Predicting Electronic Health Record Usability: Scoping Review of Adoption Models, Metrics, and Future Directions
BACKGROUND: Electronic health records (EHRs) play an essential role in modern health care, enabling data sharing and improving patient safety; however, even though vendors must adhere to International Organization for Standardization-related usability standards for EHR certification, persistent usability issues continue to undermine efficiency, contribute to clinician burden, and increase the risk of preventable errors. OBJECTIVE: This scoping review synthesizes existing research on EHR adoption and usability, emphasizing theoretical models, measurement approaches, factors, and analytic methods used to assess or predict usability. We identify gaps and opportunities for integrating predictive analytics and artificial intelligence (AI) to advance research and improve the usability of EHRs. METHODS: Following Joanna Briggs Institute and PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines, we systematically searched MEDLINE, Web of Science, IEEE Xplore, and Scopus library databases for studies published between January 1, 2009, and April 9, 2025. Inclusion criteria focused on empirical research using predictive methods or models related to EHR usability. Data were charted and synthesized thematically. RESULTS: Of the 2323 screened papers, 47 studies met inclusion criteria. Most research examined or predicted EHR adoption (not usability) using dominant frameworks, such as the technology acceptance model, unified theory of acceptance and use of technology, and the information system success model, which comprised usability. Factors related to usability-particularly perceived usefulness, perceived ease of use, effort expectancy, and facilitating conditions via the EHR adoption models-appeared frequently. Regression-based methods and structural equation modeling were the most common analytic techniques. No studies applied predictive modeling or AI to predict EHR usability. CONCLUSIONS: The focus of this study on the prediction of EHR usability and adoption for the past 15 years is distinctive in the literature. It extends prior usability reviews (mostly focusing on adoption, not prediction of usability). Predictive modeling for EHR usability remains underdeveloped throughout 2009 to 2025. Dominant frameworks in the EHR literature continue to prioritize predicting adoption over operational usability. These models rely heavily on self-reported, cross-sectional measures captured at a single postimplementation time point, embedding systematic bias and obscuring longitudinal usability dynamics. Despite the application of increasingly sophisticated predictive techniques-primarily variants of regression and structural equation modeling-usability has remained analytically subordinate to adoption and acceptance constructs for more than 15 years. As a result, widely used models, such as the technology acceptance model and unified theory of acceptance and use of technology, position usability merely as an antecedent to intention or use, rather than as an independent, system-level property that can be empirically measured, modeled, and predicted. Therefore, there is substantial opportunity to integrate predictive analytics, AI, and longitudinal usability measures to build dynamic models.
10 Apr 2026
Read appraisal →PloS one
Deep learning for adaptive chemotherapy: A DDPG-based approach to optimizing tumor-immune dynamics
In this article, we propose a deep reinforcement learning based chemotherapy regulation framework to realize personalized and dynamic optimization of cancer treatment. We use a nonlinear dynamic system to model the dynamic evolution of the tumor microenvironment including tumor cell, normal cell and immune cell interactions, with drug concentration serving as the control input variable. The Deep Deterministic Policy Gradient (DDPG) algorithm makes agents can study optimal dosing strategy in a continuous space of movement to inhibit tumor growth effectively and minimize damage to normal tissues. To make the strategy more stable, Gauss noise is added to the model to simulate physiological oscillations and uncertainties in the treatment reaction. Experimental results show that it can control the growth of tumor in various initial scenarios and the accumulation of the drug concentration with high flexibility and safety. Our technique provides a feasible technical way for precision, low toxicity adaptive chemotherapy.
9 Apr 2026
Read appraisal →JMIR cancer
Acceptance, Perceived Usefulness, and Data Sharing in Mobile Health Apps Among Patients With Breast Cancer: Cross-Sectional Survey Study
BACKGROUND: Mobile health (mHealth) apps promise to enhance patient empowerment, enable real-time health monitoring, and support self-management. Patients with breast cancer stand to benefit particularly from these capabilities, given the demanding pre- and posttreatment procedures they face. However, the effective adoption of these tools is challenged by issues including accessibility, usability limitations, and privacy concerns. OBJECTIVE: This study aimed to investigate the current usage behavior and attitudes toward mHealth apps among patients with breast cancer in Germany, focusing on acceptance, usefulness, empowerment, and data-sharing. Specifically, we aimed to identify the incentives and barriers influencing use and the intention to use mHealth apps at the time of this writing to guide their further implementation and use. METHODS: A cross-sectional online survey was conducted between November 2023 and June 2024 using a structured questionnaire developed in German. Participants with a self-reported breast cancer diagnosis were recruited via university hospital channels, patient associations, and social media. Topics included technology and health app usage, perceived usefulness, personal empowerment, and attitudes toward data sharing. The survey comprised both closed- and open-ended questions to capture quantitative patterns and qualitative insights into participants' experiences and perspectives. RESULTS: We collected responses from 90 participants with breast cancer. Participants demonstrated high engagement with digital technology; 97% (87/90) reported daily smartphone use, and 70% (63/90) owned a smartwatch or fitness tracker. Approximately 67% (60/90) of respondents had used health apps in the past year, with many relying on general fitness and lifestyle apps to support their health management. Most participants (62/90, 69%) indicated that health apps contribute to a better understanding of their personal health. Health apps were generally perceived as useful, with users highlighting their ability to provide continuous support and timely, reliable health information (n=31). This, in turn, was seen as instrumental in promoting self-management and empowerment. Respondents also showed a conditional willingness to share personal health data for research, particularly when clear benefits for treatment advancements were evident (n=82). Nonetheless, concerns were raised regarding data privacy, app usability, the need for multiple apps to cover all necessary features, and the potential increase in fear of illness (cyberchondria). CONCLUSIONS: The findings underscore the potential of mHealth apps to empower patients with breast cancer by improving health literacy and supporting self-management. To maximize their impact, future digital health tools must align closely with patient needs by incorporating adaptive, interactive features and integrating seamlessly with clinical care. Addressing key barriers-particularly data privacy and usability issues-is essential for broader adoption. Longitudinal studies are warranted to evaluate the long-term effects of health app usage on patient empowerment, quality of life, and clinical outcomes.
9 Apr 2026
Read appraisal →BMJ open
A double-blind, crossover, non-inferiority randomised controlled trial where primary care providers and patients compare human-generated and AI-generated digital health messages: the AI-CARE study protocol
INTRODUCTION: Primary care is facing multiple crises, including an increase in health misinformation. Digital health messaging by primary care providers has been shown to reach a diverse patient population. With the uptake of Generative Artificial Intelligence (GenAI) usage in healthcare, there is an important opportunity to rapidly create messages that are tailored to different populations and conditions. However, thoroughly assessing artificial intelligence (AI)-generated content is essential, as GenAI raises concerns regarding its accuracy, understandability, actionability and bias perpetuation. We aim to investigate whether digital health messages created by GenAI are evaluated as non-inferior compared with those created by human experts. METHODS AND ANALYSIS: The AI-CARE (AI to Create Accessible and Reliable patient Education materials) study is a double-blind, crossover, non-inferiority randomised controlled trial. Data collection began on 30 May 2025, and is expected to be completed at the end of May 2026. Over 12 months, 192 messages on 48 topics will be written: half by primary care and public health experts and half by a GenAI tool (OpenAI's ChatGPT). Review Panels composed of 24 primary care providers and 24 patients will evaluate these messages using an Evaluation Grid developed to assess the messages' quality of information, adaptation to the target audience, relevance and usefulness, and readiness to be shared with patients. Evaluations will be completed via online REDCap (Research Electronic Data Capture) surveys and the order in which the 192 messages appear will be randomised and will vary between individuals. Participants and analysts will be blinded to the generation source. The primary outcome will be the Clarity and Understandability score. ETHICS AND DISSEMINATION: The Research Ethics Boards of the Hôpital Montfort (24-25-11-038) and the University of Ottawa (S-12-24-11153) formally approved this study in December 2024. Reported data will be grouped and anonymised for dissemination in peer-reviewed scientific journals and conferences. TRIAL REGISTRATION NUMBER: NCT06997107.
9 Apr 2026
Read appraisal →JMIR research protocols
Enhancing Continuous Medication Safety Through e-Prescription and Clinical Decision Support Systems in Outpatient Practices and Pharmacies: Protocol for a Multiperspective Study (eRIKA Study)
BACKGROUND: Increased life expectancy is associated with increasing multimorbidity and polypharmacy, leading to a heightened risk of drug-drug interactions and adverse events, especially when multiple health care providers are involved. To address the urgent need for safer medication management in this population, tools such as medication plans (MP), electronic prescriptions (e-prescriptions), and clinical decision support systems (CDSS) offer valuable support. These instruments have the potential to enhance medication safety by providing physicians and pharmacists with a comprehensive overview of a patient's overall medication regimen and by assisting health care professionals in making informed prescribing decisions. OBJECTIVE: This study aims to improve medication therapy safety by combining e-prescriptions, the use of claims data, MPs, CDSS, and interprofessional communication. To comprehensively evaluate this complex intervention, a holistic multiphase study will be conducted, examining (1) the effectiveness of the intervention and (2) health-economic and (3) implementation-related aspects. METHODS: A multiphase study design is used. In the first phase, the intervention is implemented in selected outpatient practices (n=10) and pharmacies (n=10) in 2 regions in Germany as part of a cluster-randomized controlled trial to assess process-related outcomes. The primary outcome is the congruence between the MP and claims data. In phase 2, the intervention is scaled up in 3 regions and evaluated in a quasi-experimental study. The required sample size for the intervention group is 3528 patients, with a synthetic control group matched from existing claims data. The primary outcome is a combined end point of all-cause mortality and hospitalization within 3 months of an index prescription. Quantitative methods (descriptive, regression-based methods using claims data, calculation of the incremental cost-effectiveness ratio, and survey-based analyses of implementation-related aspects) and qualitative methods (interviews and focus groups to capture experiences of health care professionals and patients) are used. RESULTS: In phase 1, a total of 187 patients were recruited (74 in the intervention group and 113 in the control group) by June 2025. Phase 2 is currently ongoing, with data collection continuing through December 31, 2025. Final analyses are planned by March 2027. CONCLUSIONS: Medication safety in polypharmacy remains a critical challenge in Germany. This study provides multiperspective evidence supporting the nationwide implementation of the eRIKA (e-prescription as an element of interprofessional care pathways for continuous medication therapy management [eRezept als Element interprofessioneller Versorgungspfade für kontinuierliche AMTS]) intervention.
8 Apr 2026
Read appraisal →Journal of the American Heart Association
Beyond Composite Indices: Comprehensive Social Determinants Improve Heart Failure Readmission Prediction
BACKGROUND: Heart failure (HF) hospitalization readmissions are associated with a high mortality rate and strain the health care system. Both clinical factors and social determinants of health (SDOH) predict HF readmissions, but the optimal approach to incorporating area-level SDOH data remains unclear. METHODS: We merged census tract- and county-level SDOH measures with electronic health record data in a retrospective cohort of 33 579 Black and White patients with HF (Emory Healthcare, 2010-2018). Six combinations of electronic health record data with 752 area-level SDOH were evaluated using multiple machine learning models (logistic regression, random forest, XGBoost) to predict 30-day HF readmission. Models were assessed for predictive performance using area under the receiver operating characteristic curve and algorithmic fairness across race using the equalized odds ratio. RESULTS: Expanded SDOH predictor sets improved predictive performance and algorithmic fairness compared with traditional SDOH indices. The XGBoost model using expanded SDOH alongside clinical predictors provided better predictive performance (area under the receiver operating characteristic curve, 0.671) and improved algorithmic fairness across patient race (equalized odds ratio, 0.437) than models with traditional indices (area under the receiver operating characteristic curve, 0.632; equalized odds ratio, 0.329). Feature-importance analysis revealed that specific environmental predictors (housing cost burden, air quality) ranked among top predictors alongside clinical biomarkers. County-level SDOH outperformed census tract-level SDOH and matched prediction performance of clinical predictors alone. CONCLUSIONS: Including hundreds of individual SDOH indicators rather than traditional composite indices improves machine learning models' ability to predict 30-day HF readmissions. While performance gains are modest, inclusion of specific environmental factors may provide greater clinical utility and improved equity across racial groups than composite SDOH measures for guiding further research and preventive interventions.
8 Apr 2026
Read appraisal →Progress in biomedical engineering (Bristol, England)
Trustworthy AI in digital health: a comprehensive review of robustness and explainability
Ensuring trust in artificial intelligence (AI) systems is essential for the safe and ethical integration of machine learning systems into high-stakes domains such as digital health. Key dimensions, including robustness, explainability, fairness, accountability, and privacy, need to be addressed throughout the AI lifecycle, from problem formulation and data collection to model deployment and human interaction. While various contributions address different aspects of trustworthy AI, a focused synthesis on robustness and explainability, especially tailored to the healthcare context, remains limited. This review addresses that need by organizing recent advancements into an accessible framework, highlighting both technical and practical considerations. We present a structured overview of methods, challenges, and solutions, aiming to support researchers and practitioners in developing reliable and explainable AI (XAI) solutions for digital health. This review article is organized into three main parts. First, we introduce core pillars of trustworthy AI and discuss the technical and ethical challenges they pose, particularly in the context of digital health. Second, we explore application-specific trust considerations across domains such as intensive care, mental health, metabolic disease, and public health surveillance, highlighting how explainability, clinical validation, and human oversight support trust. Lastly, we present recent advancements in techniques aimed at improving robustness under data scarcity and distributional shifts, as well as XAI methods ranging from feature attribution to gradient-based interpretations and counterfactual explanations. This paper is further enriched with detailed discussions of the contributions toward robustness and explainability in digital health, the development of trustworthy AI systems in the era of large language models, and various evaluation metrics for measuring trust and related parameters such as validity, fidelity, and diversity, offering a roadmap for building safer and more reliable AI systems.
8 Apr 2026
Read appraisal →JMIR human factors
Factors Influencing User Satisfaction in Accessing Health Data: Cross-Sectional Survey of United Kingdom Adults
BACKGROUND: While patient health records (PHRs) are becoming ubiquitous, nationwide evidence on the drivers of user satisfaction in the United Kingdom remains scarce. OBJECTIVE: This study aimed to quantify determinants of PHR user satisfaction in a nationally representative sample of United Kingdom adults and to contrast real-world experiences with hypothetical expectations among individuals without PHR exposure. METHODS: We conducted a cross-sectional online survey in March 2022 using stratified quota sampling (eg, age, sex, ethnicity) through the Prolific platform. Of 1001 respondents, 533 (53%) were female (mean age 41, SD13) years, 468 (47%) reported previous PHR use ("experienced" cohort), and 533 (53%) respondents did not ("hypothetical" cohort). Primary outcomes were five satisfaction items (5-point Likert), overall PHR evaluation (0-100), and stated/anticipated PHR functions. Two-sample t-tests with unequal variances examined between-group differences; multivariate analysis of variance (MANOVA) assessed demographic equivalence. Significance was set at P<.05. RESULTS: Respondents with PHR experience rated their record easy-to-use in 79% (370/468) of cases versus an anticipated 93% (498/533) among nonusers (P<.001). Experienced users rated appointment-scheduling time-saving at 68% (319/468) compared with 72% (387/533) of nonusers (P=.17). Mean overall evaluation scores were 71.0 (SD 19.5) versus 74.2 (SD 18.6) (t₉₉₇=2.25, P=.02). MANOVA confirmed demographic balance (Wilks Λ=.99, F₇,₉₉₃=1.48, P=.18). Preferred functions across cohorts were viewing personal health information (experience 81%, hypothetical 90%) and lab results (50% vs 76%). Users lacking experience consistently over-estimated future use of carer-related functions (eg, children's data: 36% vs 12%, P<.001). CONCLUSIONS: Adults in the United Kingdom value PHRs primarily for personal data access and scheduling convenience. Nonusers displayed optimistic expectations about carer-related features and breadth of functionality, indicating an information gap. Enhancing transparency about current capabilities and expanding features that facilitate caregiving could increase satisfaction and adoption.
7 Apr 2026
Read appraisal →JMIR formative research
Performance of DeepSeek V3, DeepSeek R1, ChatGPT 4o, and ChatGPT o1 on the National Health Professional and Technical Qualification Examination (Intermediate Level) in China: Comparative Analysis
BACKGROUND: In recent years, large language models (LLMs) have undergone swift cycles of refinement and iteration. However, in the realm of clinical medicine, different LLMs' capability of logical reasoning and disease diagnosis needs further investigation. OBJECTIVE: The aim of our study was to evaluate the performance of 4 different LLMs in the National Health Professional and Technical Qualification Examination in China. METHODS: A total of 398 multiple-choice questions of 5 different question types were integrated within the examination with respect to the diagnosis or care of cases. These questions were categorized into different cardiology subspecialties and different clinical disciplines. DeepSeek V3 and R1 were accessed through an application programming interface, while ChatGPT 4o and o1 were queried via its public chat-based interface. We offered the same prompts instructing LLMs to assume the role of a physician and provide answers with explanations at the beginning of each conversation. We assessed different LLMs' performance by the accuracy in the responses to the multiple-choice questions. For the first 3 examination sections, McNemar test was used to compare the accuracy among the models, with post hoc pairwise comparisons performed using partitions of chi-square method and Bonferroni correction (significance set at P<.008). For the fourth section involving partially credit scoring, one-way ANOVA was performed to compare the mean scores among the models, with statistical significance set at P<.05. RESULTS: Both DeepSeek V3 and R1 showed superior performance in the first 3 sections of the Chinese National Health Professional and Technical Qualification Examination, achieving an overall performance of 93% and 93.6%, respectively. ChatGPT 4o and o1 achieved accuracies of 73.3% and 69%, respectively (all P<.001 compared with DeepSeek V3). For the fourth section, the performance of all 4 LLMs markedly declined compared to their results in the preceding sections. Particularly, in the section of gastroenterology and hematology, DeepSeek V3 achieved the highest accuracy, while R1 ranked first in cardiology and neurology. ChatGPT o1 achieved the highest accuracy in the topic of coronary artery disease, with no statistical significance. CONCLUSIONS: DeepSeek V3 and R1 showed remarkable potential in facilitating clinical decision-making in the Chinese professional examination, with both outperforming ChatGPT 4o and o1. Nonetheless, future research should continue evaluating their economic efficiency and susceptibility to hallucination.
7 Apr 2026
Read appraisal →JMIR human factors
Chatbot Usability Scale in Chinese Users: Cross-Cultural Adaptation and Validation Study
BACKGROUND: Chatbots are increasingly deployed across various domains; however, systematic evaluation of their usability remains limited, particularly in non-Western contexts. The 11-item Chatbot Usability Scale (BUS-11), a multidimensional instrument grounded in human-computer interaction theory, has demonstrated strong psychometric properties in prior studies; however, a validated Chinese version does not exist, despite China being one of the largest chatbot markets. OBJECTIVE: This study aimed to translate, culturally adapt, and validate the BUS-11 for Chinese users. METHODS: Following established cross-cultural adaptation procedures, the scale was forward- and back-translated, reviewed by an expert committee, and pilot-tested for clarity and feasibility. A main validation study was then conducted with 214 participants, who completed 438 evaluations of the chatbots across 10 widely used systems. RESULTS: Psychometric analyses demonstrated excellent scale-level content validity index (ie, 0.92), strong internal consistency (Cronbach α=0.92), and a clear 3-factor structure (Accessibility, Interaction Process Quality, and Information Quality), explaining 56.1% of the variance. Meanwhile, privacy or security and response time were retained as single-item indicators. The Chinese BUS-11 was concise, user-friendly, and psychometrically robust. CONCLUSIONS: This work fills a critical gap by providing the first validated instrument for assessing chatbot usability in Chinese contexts, enabling reliable cross-cultural comparisons and supporting both research and practical design evaluation in human-computer interaction.
7 Apr 2026
Read appraisal →Journal of medical Internet research
Supporting Access to Care Through Peripheral Devices and Patient-Generated Health Data: Qualitative Study
BACKGROUND: In 2016, the US Department of Veterans Affairs (VA) implemented a national initiative to distribute video-enabled tablets and peripheral devices, such as blood pressure monitors and weighing scales, to patients facing geographic, clinical, or socioeconomic challenges. Such patients could potentially benefit from health monitoring in conjunction with video-based care, as peripheral devices offer opportunities to enrich care received during a video visit and support tracking of health-related data collected outside of clinical care, or patient-generated health data. However, little is known about experiences with the devices and how they could support improved access to care. OBJECTIVE: We explored patients' experiences with VA-issued peripheral devices and their impact on video-based care and health monitoring outside of clinical visits. METHODS: We conducted in-depth semistructured interviews among patients who received VA-issued tablets and peripheral devices between 2023 and 2024. Purposive sampling was used to gather views based on gender, age, race or ethnicity, and rurality. Interviews were transcribed and analyzed using rapid qualitative analysis, guided by the Unified Theory of Acceptance and Use of Technology. RESULTS: Among 25 patients, most received a blood pressure monitor (21/25, 84%), a weight scale (14/25, 56%), and/or a pulse oximetry device (12/25, 48%). The majority reported using their peripheral devices (23/25, 92%) and tablets (19/25, 76%) to monitor their vital signs and attend video visits. Qualitative analysis yielded ten themes reflecting experiences and impacts of the devices, organized by the Unified Theory of Acceptance and Use of Technology constructs: "effort expectancy" consisted of (1) familiar and easy to use devices and (2) challenges of Bluetooth pairing and measurement; "performance expectancy" consisted of (3) integration with video visits, (4) health monitoring for peace of mind, (5) perceptions of improved vital signs and lifestyle behaviors, (6) removing obstacles to in-person care, and (7) desiring an overall picture of health; "social influence" consisted of (8) fostering care team connections and (9) promoting awareness of tablets and peripheral devices; and "facilitating conditions" consisted of (10) supportive help desk infrastructure. Overall, patients described using peripheral devices during virtual visits by syncing data to the tablet for real-time access by their care team. They also reported manually tracking and sharing patient-generated health data with their care team. Despite some challenges with Bluetooth pairing, patients found the devices easy to use and contributed to improved health and motivation. Devices also reduced logistical burdens of in-person visits, especially for those with limited mobility, visual impairments, mental health needs, or transportation barriers. CONCLUSIONS: Patients perceive that peripheral devices can enhance video-based care and support health care access and chronic disease management. Patients reported benefits to health, behavior, and communication with care teams. To maximize the impact, program enhancements should prioritize device interoperability, accessible training, and expanded outreach.
7 Apr 2026
Read appraisal →PloS one
A multi-modal data fusion and real-time monitoring on stroke risk prediction using federated learning
Predicting the risk of stroke is one of the critical problems in healthcare, which necessitates efficient solutions for providing accurate and prompt risk assessments while preserving data confidentiality. This work proposes a new framework using Federated Learning (FL) to combine Multi-Layer Perceptron (MLP) and Gated Recurrent Unit (GRU) models that are essential in analyzing multimodal data. Implemented in Python, the approach incorporates two datasets: Dataset 1, which consists of Demographic data medical history, and lifestyle data, and the second dataset, which includes the normal condition and the affected stroke condition CT scan images. Imputation of missing values, feature normalization by Min-Max scaling, and handling of imbalanced classes with SMOTE make the data pre-processing procedures exhaustive. In FL architecture three clients -Client A, Client B, and Client C - process a split multimodal dataset containing static and sequential information. Each client independently trains an MLP-GRU model. Each is applied with MLP handling static features from Dataset 1 and GRU handling sequential features from Dataset 2. To update models, Federated Averaging is used on a central server, to create a global model that is then returned to the clients for further refinement. The accuracy of the proposed method averages 99.00% and surpasses other models by 2.5% including CNN, LSTM, Random Forest, and SVM. By enhancing MLP with GRU and applying them to a privacy-preserving FL framework,The study addresses the fragmented use of multimodal medical data, where clinical records and imaging are generally evaluated separately, resulting in inadequate diagnostic support. The strategy integrates complementary modalities to create a more comprehensive perspective of patient health, enhancing healthcare predictive accuracy and decision-making. This incentive is essential for improving computational methods and linking technical advancement with medical objectives like fast diagnosis and therapy planning. The introduction emphasises the therapeutic necessity of harmonising organized and unstructured data to reduce diagnostic ambiguity. A translational approach is used to discuss how multimodal integration might improve clinical workflows, develop collaborative healthcare systems, and support sustainable medical practices. This repeated emphasis links methodological advances to real-world healthcare issues, boosting the study's academic relevance sets.
7 Apr 2026
Read appraisal →JMIR aging
AI-Enhanced Automatic Life Story Structuring for Reminiscence Therapy in Older Adults: Technical Feasibility Study
BACKGROUND: Storytelling interventions have demonstrated substantial potential in improving emotional well-being, cognitive function, and quality of life for older adults. However, its effectiveness is often limited by the challenges of processing disorganized and redundant life stories, which impose substantial cognitive demands on caregivers. Although storytelling interventions are a well-established therapeutic approach, current practices depend heavily on manual narrative organization, restricting both the scalability and consistency of treatment delivery. Prior research has primarily focused on validating the clinical outcomes of storytelling interventions, with insufficient attention given to technological solutions that could enhance narrative processing while preserving therapeutic integrity. Digital approaches to life story structuring remain underexplored, despite their potential to amplify storytelling benefits by reducing cognitive load and improving recall accuracy. OBJECTIVE: This study aims to design an event timeline generation algorithm to optimize the prior work of the Story Mosaic system. The optimized system enables (1) the automatic extraction of event elements from life narratives, (2) the automatic organization of fragmented life stories into structured timelines, and (3) the preservation of clinically relevant contextual details during compression. The goal is to reduce manual intervention costs while increasing treatment efficacy through artificial intelligence-driven narrative structuring. METHODS: We have designed a novel method, CARE event timeline (CARE-ET), which combines a temporal attention mechanism with graph-based event relationship modeling. Furthermore, we used the CARE-ET algorithm to optimize existing story collage systems. The system uses multifeature extraction technology to capture event clues from oral histories, prioritizes the 6 elements of events through a hierarchical attention mechanism, and uses adaptive compression algorithms to reduce redundancy while maintaining narrative continuity. To verify the effectiveness of the CARE-ET method, this paper adopts a multidimensional evaluation framework, which encompasses event summary assessment, timeline quality evaluation, and usability testing of the optimized system. RESULTS: The proposed CARE-ET algorithm outperforms the baseline in both narrative flow and temporal accuracy. The Story Mosaic system, optimized by the CARE-ET algorithm, underwent usability evaluation by 10 caregivers recruited for this study. Based on standardized assessment metrics, the system received an A rating for usability. The comprehensive experimental results demonstrate that the CARE-ET method can effectively structure fragmented narratives from older adults, enhancing the usability of the Story Mosaic system. CONCLUSIONS: The proposed method enables the structured extraction of representative event summaries, transforming disorganized life stories into an event timeline for caregiver-supported older adult well-being interventions. Future research should investigate longitudinal effects on cognitive preservation and explore integration with existing dementia care protocols. This work establishes a critical foundation for intelligent assistive technologies in geriatric mental health interventions.
7 Apr 2026
Read appraisal →JMIR medical informatics
Automated ICD-10-Anchored Classification of Primary Care Text Data: Development and Evaluation of a Custom Multilabel Classifier
BACKGROUND: Electronic medical records are a vast and valuable source of information, useful for tasks such as estimating disease prevalence. However, in routine primary care, much of this information is in free-text format rather than in a structured form and, therefore, not readily amenable to analysis. Manual coding of this textual data is both time-consuming and resource-intensive, making it impractical for large datasets. Although powerful open-source language models offer new opportunities for automated coding, their use on short heterogeneous primary care notes, particularly in German-language settings, remains insufficiently studied. OBJECTIVE: By providing hands-on guidance for applied health researchers, this study aims to demonstrate the effective and accurate automatic classification of free-text notes using a language model fine-tuned for automated International Statistical Classification of Diseases, Tenth Revision (ICD-10) coding. METHODS: Building on the extensive Family Medicine Research Using Electronic Medical Records (FIRE) routine database from the Institute of Primary Care at the University Hospital Zurich and the University of Zurich, we trained a large language model-based multilabel classifier on a dataset of 38,728 free-text notes, which had been manually categorized into 47 classes using specific ICD-10 codes and code ranges or nondiagnostic/ad hoc labels (eg, "unclear diagnosis," "status post"). We stratified the labeled data into training (70%), validation (15%), and posttraining test (15%) sets, ensuring similar label distributions across these sets. Using the Transformers Python library, we trained the model over 10 epochs and evaluated it on the posttraining test dataset. RESULTS: Across 48 classes, the FIRE classifier achieved strong performance on the held-out posttraining set, with F1-scores of 0.85 (micro, overall across all predictions), 0.86 (macro, mean of per-class scores treating classes equally), and 0.84 (weighted, per-class scores weighted by class frequency). CONCLUSIONS: This study demonstrates steps for training open-source large language models and highlights the potential to streamline and scale the extraction of diagnostic information for practical applications. Our model can be robustly deployed, for example, for prescreening and labeling of free-text information, thus potentially reducing the burden of repetitive and error-prone manual handling.
7 Apr 2026
Read appraisal →JMIR mental health
It Is the Journey, Not the Destination: Moving From End Points to Trajectories When Assessing Chatbot Mental Health Safety
Large language models are rapidly becoming embedded in everyday life through artificial intelligence (AI) chatbots that people use for practical assistance and companionship, as well as for support with mental health and emotional well-being. Alongside clear benefits, clinicians and public reports increasingly describe a minority of users whose interactions seem to drift over days or weeks toward strongly questionable convictions, delusions, or suicidal crises. Importantly, clinically meaningful deterioration can occur even without overtly unsafe text outputs, via more insidious processes, such as compulsive use, sleep disruption, withdrawal from human contact, and progressive narrowing of attention around the chatbot relationship. In this Viewpoint, we argue that risk often arises not at a single tipping point but through trajectory effects that accumulate across extended dialogue and that prevailing safety evaluation approaches are misaligned with this reality because they primarily score risk at discrete conversational end points often reached through scripted dialogues lasting just a single turn or several turns. Mental health benchmarks and safety suites (including clinician-informed efforts) have advanced the field by testing refusal behavior, toxicity, and adversarial prompting. However, they often treat the last message as the unit of analysis and, therefore, miss when risk-relevant relational cues, signs of validation, contradiction handling, and shifts in certainty first emerge and how they compound. We propose that mental health safety assessment should shift from end points to trajectories by (1) treating the whole dialogue, not just the end result, as the focus of evaluation; (2) reporting turn-by-turn dynamics, such as delusion confirmation and harm enablement, and timing and persistence of safety interventions; and (3) calibrating short multiturn tests against longer, clinically realistic interaction sequences that can reveal context-length effects and drift. We further argue that transcript-only evaluation is insufficient in mental health contexts. Similar language can reflect very different internal states, and the relationship between expressed psychopathology and real-world harm is nonlinear. Therefore, safety research should incorporate proximal human outcomes following interactions (eg, shifts in certainty, openness to counterevidence, arousal, urge to continue, and subsequent sleep or behavior) and build a prospective clinical surveillance infrastructure that supports transcript donation with consent and linkage to health outcomes. Together, these steps would enable benchmarks that are clinically relevant and better aligned with the types of harms now being observed in real-world chatbot use.
7 Apr 2026
Read appraisal →Lifetime data analysis
Q-learning via deep learning-based Buckley-James method for non-linear censored data
In healthcare, personalized treatment strategies are vital for improving patient outcomes, especially under right-censored survival data. We propose Dynamic Deep Buckley-James Q-Learning, a novel counterfactual Q-learning algorithm that integrates deep learning with the Buckley-James method to simultaneously address censoring and nonlinear modeling challenges. By explicitly capturing complex, nonlinear interactions between covariates and treatment effects, the algorithm robustly estimates optimal dynamic treatment regimes. Leveraging a counterfactual framework, we define and estimate potential survival outcomes under hypothetical treatment sequences, enabling unbiased Q-function estimation even in the presence of time-dependent covariates and right censoring. The algorithm maximizes the expected imputed survival reward under these counterfactual scenarios. Simulation studies and real-world data analysis demonstrate its superior performance in predictive accuracy and treatment decision-making, offering a powerful framework for individualized care in complex clinical settings.
6 Apr 2026
Read appraisal →PloS one
IOTA Tangle-based traceability framework for wheat crop supply chain
Distributed ledger technology (DLT) has emerged as a transformative solution across industries, offering decentralization, transparency, and security. Among DLT architectures, IOTA's Tangle stands out as a scalable, feeless framework well-suited for supply chain applications, particularly in critical sectors like agriculture. This work leverages IOTA's Tangle to design a secure and traceable wheat supply chain system, addressing challenges such as opacity, fraud, and inefficiency. The proposed solution is an IOTA-based framework designed for enhancing supply chain traceability of wheat, offering trustworthy, transparent, secure and scalable services. Based on smart contracts, a transaction processing system was developed to enhance transparency and traceability and automate tasks. Furthermore, we used various technologies for increased visibility and operational efficiency throughout the supply chain, including IPFS (InterPlanetary File System) for storing extensive data such as stakeholder certifications, AI (Artificial Intelligence) for predicting crop yields, and Self-Sovereign Identity (SSI) for enhanced security. In addition, a government-provided mobile app helps silo managers and police officers verify transportation credentials and prevent contraband. The implementation of this system is expected to enhance supply chain resilience and transparency, particularly in Tunisia, thereby supporting the country's goal of food autonomy.
6 Apr 2026
Read appraisal →European journal of cardiovascular nursing
A review and promise of digital health technologies to promote global cardiovascular health
Digital health technologies include an array of mobile or digital devices and apps that may be used by patients to support lifestyle modification or medical apps that monitor vital parameters and provide data that may influence treatment decisions. The use of digital health technologies is a growing opportunity for disease prevention and management of cardiovascular disease (CVD) for both patients to engage in self-management as well as nurses to monitor patients remotely and facilitate clinical decision support. The aims of this review article are to provide an overview of pertinent digital health technologies in CVD care, address social determinants of health and other enablers/barriers to digital health implementation, and provide perspectives on novel models of care incorporating digital technologies. Based on the latest evidence, recommendations for special considerations and implementation of digital health technologies in low- and middle- vs high-income countries are presented. In addition, tables reviewing functionality of the digital tools and global examples of their use are provided. Extensive empirical evidence from diverse populations suggests adoption of these innovative tools has been low in clinical settings. We provide a table summarizing these barriers and include the facilitators that can improve the adoption of digital health tools to improve behavioral lifestyle modification. For learning purpose, implementation of an innovative model of care is illustrated graphically and in text. We include a call to action to achieve the maximum impact of digital technology to support nurses who work in almost all settings improving CVD prevention and management globally. The multiple roles of nurses using digital health for global CVD prevention and management are discussed in the settings of clinical practice, research, education, public health, health policy/leadership, and informatics.
5 Apr 2026
Read appraisal →PloS one
A robust deep learning approach for impulse noise filtering using hybrid auto-encoder with fuzzy median filter
De-noising convolutional neural networks (DnCNNs), are a powerful nonlinear mapping models in image processing for impulse noise removal. During training and validation, a set of 12 standard testing images is used to evaluate model performance. DnCNNs demonstrate strong capability in classification of impulse noise with excellent results. To evaluate de-noising performance, a suitable noise ratio should be added so that most appropriate DnCNN model can be used for impulse noise detection. This research proposes an effective image restoration technique that integrates DnCNN and an autoencoder with a fuzzy median filter to detect and eliminate high-density impulse noise. The proposed deep learning de-noising technique used to classify noisy and clean pixels, and result are presented in different metrics such as accuracy, FPR, FNR and f1 score. Further, to remove impulse noise an auto-encoder with fuzzy median filter are used that then reconstructs the clean image based with parametric values. Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), are used to assess our methodology, it is compared to conventional impulse noise filtering techniques, experimental results indicate a significant improvement in image quality. Based on the final de-noised images, this research contributes to developing deep learning-based, de-noising techniques that enhance image restoration quality while preserving image details and essential features.
5 Apr 2026
Read appraisal →Cancer causes & control : CCC
Comparing endocrine therapy adherence across EHR and claims data by race and ethnicity, language, and insurance.
PURPOSE: We compared adherence measurements across electronic health records (EHRs) and claims data for 79 HR + breast cancer survivors, considering their race, ethnicity, language, and insurance status. METHODS: Adherence was evaluated using the medication possession ratio (MPR) and proportion of days covered (PDC). The correlation between EHR and claims data was assessed with Spearman correlation coefficients, while Kaplan-Meier curves were employed to visualize and analyze the time to the first instance of non-adherence and the time to discontinuation. RESULTS: A total of 75 breast cancer patients receiving care at Massachusetts General Hospital (MGH) were identified. While there was a positive correlation, EHR data tended to overestimate adherence based on medication prescriptions, while claims data captured adherence through prescription fills. Kaplan-Meier analyses demonstrated differences in the time to first non-adherence between EHR and claims data sources, with observed variations based on language and insurance status. CONCLUSION: Our findings suggest that EHR and claims data offer complementary insights into adherence measurement, and their integration could improve adherence assessment across diverse populations.
5 Apr 2026
Read appraisal →American journal of human genetics
The Electronic Medical Records and Genomics study: Design and analytic framework for assessing the impact of genome-informed risk assessments.
The Electronic Medical Records and Genomics (eMERGE) Network developed and implemented a genome-informed risk assessment (GIRA) to communicate genomic (polygenic risk scores [PRSs], integrated risk scores [IRSs], and monogenic results), clinical, and family history-based risk for 11 chronic diseases and provide recommended healthcare recommendations. GIRA reports have now been returned to 23,840 participants and their providers in a large prospective cohort study. We present here the study design and analysis framework for assessing the attributable impact of GIRA return. Pre-specified outcomes include (1) provider/participant adoption of recommended healthcare actions, (2) new diagnosis of disease, (3) treatment initiation/intensification, and (4) clinical outcomes (surrogate markers or clinical events). We assess outcomes in high risk vs. not-high-risk participants, adjusting for covariates. We evaluate the effect of PRS/IRS at pre-established high-risk thresholds using regression discontinuity (RD), a quasi-experimental method that mimics randomization near a cutoff, enabling estimation of causal effects and controlling for unobserved confounders. Monogenic and family history-based risk stratification are analyzed using logistic regression. With 23,840 participants and 12 months of follow-up, the study is powered to detect differences of 2%-11% with 80% power (α = 0.05 in the adoption outcome). Longer follow-up will be required to enable assessment of new disease diagnosis, treatment changes, and clinical outcomes. Through innovative RD analyses and defined outcomes and comparison groups, this study will provide new insights into the real-world clinical impact of genomic risk assessment, address critical evidence gaps, advance understanding of genomic medicine outcomes, and inform future research.
5 Apr 2026
Read appraisal →BMJ open
Synthetic data-augmented machine learning for 30-day readmission prediction in patients with chronic conditions: a retrospective real-world study
OBJECTIVES: To develop and evaluate an explainable machine learning framework enhanced with synthetic data generation to predict unplanned 30-day hospital readmissions among patients with chronic obstructive pulmonary disease (COPD), heart failure (HF) and type 2 diabetes mellitus (T2DM), and to identify key clinical and social predictors of readmission. DESIGN: A retrospective cohort study using electronic health record data incorporating both structured variables and information extracted from unstructured clinical notes. Synthetic data were generated using advanced resampling and deep learning-based techniques to address outcome imbalance and improve model training. SETTING: Intensive care unit and general ward admissions at a single tertiary academic medical centre included in the MIMIC-IV (Medical Information Mart for Intensive Care IV) database. PARTICIPANTS: Adult patients (≥18 years) were admitted with a primary diagnosis of COPD (n=14 050), HF (n=7097) or T2DM (n=12 735) between 2008 and 2019, with complete 30-day follow-up and no in-hospital mortality during the index admission. PRIMARY AND SECONDARY OUTCOME MEASURES: The primary outcome was unplanned all-cause hospital readmission within 30-days of discharge. Predictors were drawn from six domains, including demographics, comorbidities, clinical acuity, therapies, behavioural factors and care continuity. Predictive performance was evaluated using multiple machine learning methods and fivefold cross-validation, with model interpretability assessed using established goal and local explanation approaches. RESULTS: Ensemble-based machine learning models demonstrated the strongest predictive performance across all three disease cohorts. Key predictors of readmission included higher illness severity, greater comorbidity burden, medication non-adherence, gaps in preventive care and limited social support. Models incorporating synthetic data augmentation showed improved discrimination compared with models trained on original data alone. CONCLUSIONS: An explainable synthetic-data driven framework incorporating clinical, behavioural and social data can support prediction of 30-day readmissions among patients with common chronic conditions using routinely available electronic health record data.
5 Apr 2026
Read appraisal →The International journal of oral & maxillofacial implants
Accuracy and Reliability of Artificial Intelligence Chatbots as Public Information Sources in Implant Dentistry
PURPOSE: To evaluate the accuracy, completeness, comprehensibility, and reliability of widely available artificial intelligence (AI) chatbots when addressing clinically significant queries pertaining to implant dentistry. MATERIALS AND METHODS: A total of 20 questions were devised based on a compiled list that were most frequently asked or encountered during patient consultations by three experienced prosthodontists. These questions were asked to ChatGPT-3.5, Gemini, and Copilot AI chatbots on three separate occasions in 12-day intervals. A three-point Likert scale (0 = incorrect; 1 = incomplete or partially correct; 2 = correct) and a two-point scale (true and false) were employed by the authors to grade the accuracy of the responses independently. Completeness and comprehensibility were also evaluated using a three-point Likert scale (scored 1, 2, or 3, with lower scores indicating a worse result). In addition to the questions generated by the specialists, AI chatbots were asked to provide five frequently asked questions and responses regarding dental implant prostheses. A comparison of total scores obtained from the chatbots was made with one-way analysis of variance (ANOVA). Twopoint scale data were analyzed via chi-square test. The reliability of the responses for each chatbot was analyzed by assessing the consistency of repeated responses via calculating Cronbach's alpha (α) coefficients. RESULTS: When the total accuracy scores of the chatbots were analyzed out of a 40-point total score (ChatGPT-3.5 = 28.78 ± 4.06; Gemini = 30.89 ± 4.08; Copilot = 29.11 ± 3.22), one-way ANOVA revealed no statistically significant differences (P = .461). For the evaluation of two-point accuracy scale data, which were analyzed via chi-square test, no statistical differences were revealed among the chatbots (P = .336). Gemini showed a higher completeness level than ChatGPT-3.5 (P = .011). There was no statistically significant difference among AI chatbots in terms of comprehensibility. Copilot demonstrated the greatest overall consistency among the three chatbots, with a Cronbach's α value of .863. This was followed by ChatGPT-3.5 with a Cronbach's α value of .779 and Gemini with a Cronbach's α value of .636. CONCLUSIONS: The accuracy of all three chatbots were found to be similar. All three chatbots demonstrated an acceptable level of consistency. However, given the low accuracy rate of chatbots in answering questions, it is clear that they should not be relied upon as a decision-maker, and the clinician's opinion must be given priority.
4 Apr 2026
Read appraisal →JMIR mental health
AI Chatbots for Mental Health Self-Management: Lived Experience-Centered Qualitative Study
BACKGROUND: Large language models (LLMs) now enable chatbots to engage in sensitive mental health conversations, including depression self-management. Yet their rapid deployment often overlooks how well these tools align with the priorities of people with lived experiences, which can introduce harms such as inaccurate information, lack of empathy, or inadequate crisis support. OBJECTIVE: This study explores how people with lived experience of depression experience an LLM-based mental health chatbot in self-management contexts, and what perceived benefits, limitations, and concerns inform harm-mitigating design implications. METHODS: We developed a technology probe (a GPT-4o-based chatbot named Zenny) designed to simulate depression self-management scenarios grounded in prior research. We conducted interviews with 17 individuals with lived experiences of depression, who interacted with Zenny during the session. We applied qualitative content analysis to interview transcripts, notes, and chat logs using sensitizing concepts related to values and harms. RESULTS: We identified 3 themes shaping participants' evaluations: (1) informational accuracy and applicability, including concerns about incorrect or misleading information, vagueness, and fit with personal constraints; (2) emotional support vs need for human connection, including validation and a judgment-free space alongside perceived limits of machine empathy; and (3) a personalization-privacy dilemma, where participants wanted more tailored guidance while withholding sensitive information and using privacy-preserving tactics. CONCLUSIONS: People with lived experience of depression evaluated LLM-based mental health chatbots through intertwined priorities of actionable information, emotional validation with clear limits, and personalization that does not require unsafe data disclosure. These findings suggest concrete design strategies to mitigate harms and support LLM-based tools as complements to, rather than replacements for, human support and recovery.
4 Apr 2026
Read appraisal →Regional anesthesia and pain medicine
Human versus artificial intelligence: evaluating ChatGPT's performance in conducting published systematic reviews with meta-analysis in chronic pain research
INTRODUCTION: Artificial intelligence (AI), particularly large-language models like Chat Generative Pre-Trained Transformer (ChatGPT), has demonstrated potential in streamlining research methodologies. Systematic reviews and meta-analyses, often considered the pinnacle of evidence-based medicine, are inherently time-intensive and demand meticulous planning, rigorous data extraction, thorough analysis, and careful synthesis. Despite promising applications of AI, its utility in conducting systematic reviews with meta-analysis remains unclear. This study evaluated ChatGPT's accuracy in conducting key tasks of a systematic review with meta-analysis. METHODS: This validation study used data from a published meta-analysis on emotional functioning after spinal cord stimulation. ChatGPT-4o performed title/abstract screening, full-text study selection, and data pooling for this systematic review with meta-analysis. Comparisons were made against human-executed steps, which were considered the gold standard. Outcomes of interest included accuracy, sensitivity, specificity, positive predictive value, and negative predictive value for screening and full-text review tasks. We also assessed for discrepancies in pooled effect estimates and forest plot generation. RESULTS: For title and abstract screening, ChatGPT achieved an accuracy of 70.4%, sensitivity of 54.9%, and specificity of 80.1%. In the full-text screening phase, accuracy was 68.4%, sensitivity 75.6%, and specificity 66.8%. ChatGPT successfully pooled data for five forest plots, achieving 100% accuracy in calculating pooled mean differences, 95% CIs, and heterogeneity estimates (I2 score and tau-squared values) for most outcomes, with minor discrepancies in tau-squared values (range 0.01-0.05). Forest plots showed no significant discrepancies. CONCLUSION: ChatGPT demonstrates modest to moderate accuracy in screening and study selection tasks, but performs well in data pooling and meta-analytic calculations. These findings underscore the potential of AI to augment systematic review methodologies, while also emphasizing the need for human oversight to ensure accuracy and integrity in research workflows.
4 Apr 2026
Read appraisal →The American journal of gastroenterology
A Large Language Model Assistant for Summarizing Hepatology Referral Documents
INTRODUCTION: To use a large language model (LLM) to create accurate and useable summaries of patient records for clinicians triaging new hepatology referrals. METHODS: We developed a comprehensive list of data elements required to triage a new hepatology referral and engaged in an iterative prompt engineering process to instruct an LLM to extract relevant data from patient referral documents. The final prompt was used on 50 original patient records from June to July 2025 to generate corresponding artificial intelligence (AI) summaries, which were assigned to 2 providers to review and triage according to their usual process. We assessed time to triage original vs AI files and accuracy of the AI files. A linear mixed-effects model was used to determine an adjusted time ratio comparing the time to triage AI files vs original files. RESULTS: AI-generated summaries were significantly shorter than original files (median [interquartile range] 2 [2-3] vs 23 [10.2-38.8] pages, P < 0.001). AI summaries had high accuracy (median [interquartile range]: 94.6% [86.5%-97.3%]) with a low hallucination rate. Use of the AI summaries led to a 60% reduction in triage time (adjusted mean triage time of 37.2 seconds for AI files vs 94.2 seconds for original files, P < 0.001). DISCUSSION: The use of an LLM led to significantly reduced document length, maintained an appropriate level of accuracy, and led to a significant decrease in clinician time to review the patient record. Future steps involve creating a fully automated workflow that is integrated into the electronic health record for widespread use.
4 Apr 2026
Read appraisal →Angiology
Artificial Intelligence Techniques for Prognostic and Diagnostic Assessments in Peripheral Artery Disease: A Scoping Review
Peripheral artery disease (PAD) is a major public health concern worldwide, associated with high risk of mortality and morbidity related to cardiovascular and adverse limb events. Despite significant advances in both medical and interventional therapies, PAD often remains under-diagnosed, and the prognosis of patients can be difficult to predict. Artificial intelligence (AI) has brought a wide range of opportunities to improve the management of cardiovascular diseases, from advanced imaging analysis to machine-learning (ML)-based predictive models, and medical data management using natural language processing (NLP). The aim of this review is to summarize and discuss current techniques based on AI that have been proposed for the diagnosis and the evaluation of the prognosis in patients with PAD. The review focused on clinical studies that proposed AI-methods for the detection and the classification of PAD as well as studies that used AI-models to predict outcomes of patients. Through evaluation of study design, we discuss model choices including variability in dataset inputs, model complexity, interpretability, and challenges linked to performance metrics used. In the light of the results, we discuss potential interest for clinical decision support and highlight future directions for research and clinical practice.
4 Apr 2026
Read appraisal →BMJ open
Use of ambient AI scribe in physicians' clinical documentation: a protocol for a systematic review on effectiveness, efficiency, and satisfaction
INTRODUCTION: Clinical documentation is a significant driver of burnout among physicians. Ambient artificial intelligence (AI) scribes, which leverage generative large language models to automate the creation of clinical notes from patient-physician conversations, are rapidly emerging as a potential solution. While these tools promise to enhance efficiency and reduce administrative tasks, concerns about the quality, accuracy and potential biases persist. There is now a need for a systematic synthesis of evidence to evaluate the impact of these technologies in clinical practice. To assess the effects of ambient AI scribes on physicians' clinical documentation, the specific objectives are to: (1) evaluate the effectiveness of these tools on documentation, including accuracy and completeness; (2) synthesise evidence on the impact on physician efficiency after adoption, including time spent on documentation and (3) examine physicians' satisfaction with these tools, including physicians' perceived burden. METHODS AND ANALYSIS: A systematic review of quantitative or mixed-method studies as well as preprints will be conducted. We will perform a comprehensive search of four electronic databases (PubMed, IEEE Xplore, APA PsycInfo and Web of Science, along with medRix and ClinicalTrials.gov for preprints) for empirical studies published between January 2023 and March 2026. The review will synthesise studies comparing physicians' use of ambient AI scribes with traditional documentation approaches. Given the anticipated heterogeneity of the studies, a narrative synthesis will be employed to summarise the findings. Where common quantitative outcomes exist, effect sizes will be calculated using Hedges' g, mean differences or risk ratios/odds ratios as appropriate. The overall quality of evidence will be assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework. ETHICS AND DISSEMINATION: As no patient data are involved in the data collection, no ethical approval is acquired. Results will be disseminated in a peer-reviewed, open-access journal, and presented at relevant academic conferences. PROSPERO REGISTRATION NUMBER: CRD420251149086.
4 Apr 2026
Read appraisal →Current opinion in anaesthesiology
Artificial intelligence-driven predictive analytics for postoperative management and recovery in trauma patients
PURPOSE OF REVIEW: Post-traumatic care is evolving from a reactive, protocol-driven paradigm to a predictive, personalized approach. This review examines how artificial intelligence and machine learning are redefining postoperative management by predicting complications before they manifest. RECENT FINDINGS: Recent literature (2023-2025) highlights three major advances: (a) the validation of gradient boosting algorithms (e.g. Extreme Gradient Boosting) that significantly outperform traditional scoring systems for predicting trauma-induced coagulopathy; (b) the development of interpretable, phenotype-specific models for venous thromboembolism risk stratification, particularly in traumatic brain injury; and (c) the emergence of real-time sepsis prediction tools that account for the sterile inflammation inherent to trauma. However, a recurring limitation in current research is the reliance on retrospective datasets and single-center validations, underscoring the critical need for rigorous external validation across diverse patient populations before widespread clinical adoption. SUMMARY: Artificial intelligence is not merely a monitoring tool but a driver of precision medicine in trauma. By leveraging diverse modalities, from computer vision in radiology to natural language processing in electronic health records, clinicians can now anticipate adverse events. To bridge the gap between algorithm and bedside, future efforts must focus on overcoming significant implementation barriers, such as data interoperability, and ensuring model generalizability.
4 Apr 2026
Read appraisal →Current opinion in anaesthesiology
Data integration and systems interoperability: the prerequisite for artificial intelligence in anesthesiology
PURPOSE OF REVIEW: Anesthesiology generates large volumes of heterogeneous perioperative data, including high-resolution physiological signals, clinical documentation, and device-generated information. Despite this richness, the clinical deployment of artificial intelligence systems remains limited. This review examines how limitations in data integration and systems interoperability constrain the translation of artificial intelligence into routine practice. RECENT FINDINGS: Recent literature indicates that the principal barrier to artificial intelligence adoption in anesthesiology is not algorithmic performance but inadequate data integration. Data are distributed across anesthesia information management systems, electronic health records, and multiple medical devices, with variable data models, inconsistent semantic representations, and limited temporal synchronization. These systems were primarily designed for documentation, billing, and medicolegal purposes, rather than real-time analytics or secondary data use. In contrast, intensive care medicine has benefited from early investments in shared databases, which have facilitated reproducible artificial intelligence research and validation. Although emerging standards such as HL7 Fast Healthcare Interoperability Resources and common data models provide technical frameworks for interoperability, their implementation in anesthesiology remains partial. SUMMARY: The effective deployment of artificial intelligence in anesthesiology depends on the development of interoperable, high-quality data infrastructures. Establishing standardized data models, semantic harmonization, temporal alignment, and robust data governance is a prerequisite for scalable, trustworthy artificial intelligence-enabled perioperative care.
4 Apr 2026
Read appraisal →Patient education and counseling
Patient and clinician engagement with generative artificial intelligence (GenAI): A scoping review of implications for patient-centered communication
OBJECTIVE: To examine the influence of the emerging use of generative artificial intelligence (GenAI) within electronic health records and among the public on the patient-centeredness of communication in healthcare. METHOD: In this scoping review, we conducted a systematic search for peer-reviewed studies in PubMed and PsycInfo that empirically examined GenAI involvement in clinical communication. We then mapped study findings onto a well-established framework for patient-centered communication. RESULTS: Our search yielded 67 studies for analysis. Results suggest that integration of GenAI into healthcare communication has the potential to increase clinician efficiency in interacting with patients, to expand channels for patients to obtain information about their healthcare, and to enhance empathy in clinical communication. However, findings also indicate variability in the quality of information produced by GenAI, the potential for GenAI to recast the clinician as a technical supervisor rather than a humanistic care provider, and several issues of equity and privacy raised by engagement with GenAI. CONCLUSION: As GenAI becomes more prevalent in healthcare, rigorous examination of GenAI is needed to ensure that its development and implementation aids rather than hinders patient-centered communication. We conclude with an agenda for further research on GenAI grounded in the PCC framework underlying our review. PRACTICE IMPLICATIONS: Findings from this review highlight the current potential benefits and limitations of GenAI as a third party to clinical communication. Continued efforts toward developing and applying GenAI for effective healthcare communication should focus on protecting patients from potential drawbacks and maximizing nascent benefits for patient-centered communication.
4 Apr 2026
Read appraisal →International journal of medical informatics
Automated extraction of fluoropyrimidine treatment and treatment-related toxicities from clinical notes using natural language processing.
OBJECTIVE: Fluoropyrimidines are widely prescribed for colorectal and breast cancers, but are associated with toxicities such as hand-foot syndrome and cardiotoxicity. Since toxicity documentation is often embedded in clinical notes, we aimed to develop and evaluate natural language processing (NLP) methods to extract treatment and toxicity information. MATERIALS AND METHODS: We constructed a gold-standard dataset of 236 clinical notes from 204,165 adult oncology patients. Domain experts annotated categories related to treatment regimens and toxicities. We developed rule-based, machine learning-based (Random Forest [RF], Support Vector Machine [SVM], Logistic Regression [LR]), deep learning-based (BERT, ClinicalBERT), and large language models (LLM)-based NLP approaches (zero-shot and error analysis prompting). A 5-fold cross validation were conducted to validate each model. RESULTS: Error analysis prompting achieved optimal precision, recall, and F1 scores for treatment (F1 = 1.000) and toxicities extraction (F1 = 0.965), whereas zero-shot perform moderately (treatment F1 = 0.889, toxicities extraction F1 = 0.854) Rule-based reached F1 = 1.000 for treatment and F1 = 0.904 for toxicities extraction. LR and SVM ranked second and fourth for toxicities extraction (LR F1 = 0.914, SVM F1 = 0.903). Deep learning and RF underperformed, with performance of BERT reached F1 = 0.792 for treatment and F1 = 0.837 for toxicities extraction.,ClinicalBERT reached F1 = 0.797 for treatment and F1 = 0.884 for toxicities extraction). RF reached F1 = 0.745 for treatment and F1 = 0.853 for toxicities extraction. DISCUSSION: LMM-based error analysis outperformed all others, followed by machine learning methods. Machine learning and deep learning methods were limited by small training data and showed limited generalizability, particularly for rare categories. CONCLUSION: LLM-based error analysis most effectively extracted fluoropyrimidine treatment and toxicity information from clinical notes, and has strong potential to support oncology research and pharmacovigilance.
4 Apr 2026
Read appraisal →Journal of biomedical informatics
Natural language generation in healthcare: A review of methods and applications
OBJECTIVE: This study presents a systematic review of natural language generation (NLG) methods and applications in the medical domain, providing quantitative and qualitative analyses to answer four key research questions regarding methods, evaluation, applications, and challenges of NLG in healthcare. METHODS: We searched PubMed, ACM Digital Library, Web of Science, Science Direct, Scopus, Embase, and ACL Anthology for NLG-related studies in healthcare from 2018 to 2024. Out of 3,988 research articles, 113 met the inclusion criteria and were analyzed across data modality, model architecture, evaluation metrics, and application domain. RESULTS AND CONCLUSION: NLG in healthcare has grown substantially, with annual publications increasing from 2 in 2018 to 40 in 2024. Of the 113 included studies, text-to-text generation was the most common data modality (65.5%), followed by image-to-text (19.5%) and multimodal-to-text (15.0%). Transformer-based architectures were dominant, especially encoder-decoder models (61.6%). Automatic evaluation metrics such as ROUGE (81.4%) and BLEU (57.5%) were widely used. Human evaluation metrics, such as Likert scales (31.9%), were increasingly adopted. The four most prevalent application domains include summarization (e.g., discharge summaries, radiology reports), clinical documentation, medical dialogue, and data augmentation. The transformer-based large language models (LLMs) and the accumulation of large-scale multimodal clinical datasets have remarkably advanced NLG in healthcare. However, challenges remain in factual consistency, explainability, evaluation robustness, and AI safety. Addressing these challenges is essential for the adoption of NLG in various healthcare applications.
4 Apr 2026
Read appraisal →Annals of plastic surgery
Predicting Surgical Outcomes in Breast Reconstruction With Machine Learning: A Systematic Review
INTRODUCTION: The applications of artificial intelligence (AI) in plastic surgery have grown considerably in recent years. As large patient datasets become more accessible, surgeons are increasingly leveraging machine learning (ML), a subset of AI, to predict patient outcomes and guide surgical decision-making. This review evaluates the relative performance of ML prediction models in breast reconstruction. METHODS: A systematic review was conducted utilizing PubMed, Scopus, and EMBASE according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Studies using ML to predict patient outcomes in breast reconstruction were included. The type of ML model and the specific outcome measures were reported. Performance of the models was reported as area under the receiver operating characteristic curve and compared using descriptive statistics, multivariate linear regression, and random-effects meta-regression in RStudio. RESULTS: Our search yielded 1025 citations, of which 24 were assessed for eligibility. Fourteen studies met the inclusion criteria and were sought for data extraction. There were 19 ML models and 11,013 patients assessed across 92 testing conditions. Models were trained on varying patient demographics, comorbidities, and operative characteristics, whereas outcomes assessed included various surgical complications or patient satisfaction using BREAST-Q. The median area under the receiver operating characteristic curve of all models was 0.71 (interquartile range = 0.16). When adjusting for the number of patients, number of predictors, ML model category, and outcome of interest, models predicting BREAST-Q performed higher with skin necrosis used as the reference outcome (β = 0.13, P < 0.01). After adjusting for number of patients and predictors, models that employed strategies to mitigate class imbalance were associated with higher model discrimination (β = 0.038; 95% CI, 0.002-0.075; P = 0.041). CONCLUSION: Machine learning applications for risk prediction and surgical planning are growing rapidly. The models evaluated in this review demonstrated the ability to predict a variety of outcomes, with models predicting BREAST-Q, various surgical outcomes, and those reporting class imbalance methods leading to higher model discrimination. Notably, covariate adjustment and study heterogeneity may have impacted these associations. As ML models are increasingly integrated into plastic surgery practice, standardized reporting practices are essential to promote reproducibility and cross-study comparison.
4 Apr 2026
Read appraisal →Stroke
Balancing Innovation and Responsibility: Ethical and Privacy Challenges in Stroke Digital Health
Digital health solutions are being increasingly used to support care delivery across various medical fields. In stroke care, technology has contributed to enhancing prehospital workflows, facilitating communication among health care professionals, and enabling fast-track systems for hyperacute stroke management. Furthermore, digital tools are used to improve etiological diagnosis and allow for more tailored long-term follow-up of stroke patient survivors. Digital technologies have become an integral part of stroke management, and the rapid evolution of artificial intelligence continues to accelerate and drive this transformation. However, ethical and privacy considerations are central and represent a major challenge for the design and implementation of digital solutions in health care. Health care professionals involved in stroke care require a clear understanding of the digital tools they use, including their potential, limitations, and biases, and must be aware of the responsibilities inherent in relying on such systems for treatment decisions and outcome prediction. This topical review aims to examine current digital innovations in stroke care, with a particular focus on the key ethical and privacy challenges they raise.
4 Apr 2026
Read appraisal →PLoS medicine
Machine learning assisted differentiation of low acuity patients at dispatch: The MADLAD randomized controlled trial
BACKGROUND: Resource Constrained Situations (RCS) at Emergency Medical Dispatch centers where there are more patients requiring an ambulance than there are available ambulances are common. Machine Learning (ML) techniques offer a promising but largely untested approach to assessing relative risks among these patients. The study aims to establish whether the provision of ML-based risk scores predicting patient outcomes improves the ability of dispatchers to identify patients at high risk for deterioration in RCS and dispatch the first available ambulance to them. METHODS AND FINDINGS: We performed a parallel-group, randomized trial of adult patients assessed by a dispatch nurse at two study sites in Sweden as requiring a low-priority ambulance response in RCS. Patients were randomized 1:1 to be prioritized with the aid of an ML-based risk assessment tool, or per current clinical practice. The primary outcome was defined in terms of whether the first available ambulance was sent to the patient with the highest National Early Warning Score (NEWS 2) based on subsequently collected vital signs. A total of 1,245 RCS were included in the study. In the intervention arm, 68.3% of RCS were assessed correctly per the primary outcome versus 62.5% in the control group, corresponding to an odds ratio of 1.28 (95% CI [1.00, 1.63], p = 0.047). This study was limited to only patients determined to require a low-priority ambulance response in two Swedish regions, and was underpowered for the primary outcome due to a smaller than expected sample size. CONCLUSION: This study suggests that clinical ML-based decision support tools may have the ability to influence care provider decisions and improve their capacity to rapidly differentiate between high- and low-risk patients at dispatch. Further research should establish the suitability of these tools in larger cohorts, for patients with both higher- and lower-levels of priority, and in other settings. The trial was registered at ClinicalTrials.gov (NCT04757194).
4 Apr 2026
Read appraisal →Annals of oncology : official journal of the European Society for Medical Oncology
Next-generation multicenter studies: using artificial intelligence to automatically process unstructured health records of patients with lung cancer across multiple institutions
BACKGROUND: Manual abstraction of real-world data (RWD) from unstructured health records (HRs) remains resource intensive, error prone, and highly variable across institutions. Large language models (LLMs) offer a scalable alternative, but their performance in multicenter oncology settings is not fully validated. PATIENTS AND METHODS: We conducted a multicenter study within the French Large & Unified Cancer Cohort (LUCC) consortium to compare the accuracy of artificial intelligence (AI)-based data extraction against manual abstraction by clinical research professionals. A fine-tuned LLM was applied to de-identified unstructured HRs in PDF format to extract 31 variables from lung cancer patients across 10 centers. Ground truth was defined as concordant values across sources, with discrepant cases adjudicated by a blinded expert. The primary endpoint was the extraction error rates. Secondary endpoints included per-variable performance, interinstitutional variability, F1-score for multiple-choice variables, added value of hybrid AI-human workflows, and survival analyses. RESULTS: Among 10 327 patients with AI-based extraction, 311 were included in the test cohort. Across 8708 datapoints for 28 variables with only one correct answer, the LLM achieved a 7.0% error rate, outperforming manual abstraction (14.2%, P <0.001). The F1-scores of three multiple-choice variables were superior (gene alterations 0.97 versus 0.86, comorbidities 0.86 versus 0.76, metastatic sites 0.71 versus 0.69). Interinstitutional variance was lower with AI (0.12% versus 0.39%). A hybrid approach with targeted human review of 30% of low-confidence AI outputs further decreased error rates to 4.4%. Survival analyses based on AI-extracted data closely matched ground truth, with similar median overall and progression-free survival. CONCLUSIONS: In a multicenter setting, our AI pipeline yielded lower error rates and greater consistency than manual abstraction. These findings support the feasibility of next-generation, AI-enabled multicenter studies to generate high-quality RWD at scale, with potential applicability in prospective clinical trials.
4 Apr 2026
Read appraisal →Respiratory medicine
A review of artificial intelligence-based research on chronic obstructive pulmonary disease.
In recent years, with the rapid development of artificial intelligence (AI), Chronic Obstructive Pulmonary Disease (COPD), one of the world's three major chronic diseases, has achieved remarkable progress in diagnosis, grading, and prognosis, which is of great significance for promoting the clinical transformation of respiratory diseases. To deeply explore the application of AI in the diagnosis and management of COPD, this paper reviews recent studies based on machine learning and deep learning, covering screening and diagnosis, disease grading and assessment, disease management and monitoring, and treatment. First, the technical basis of COPD-related research is analyzed from five perspectives: traditional research methods, supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. Then, the commonly used datasets and model evaluation metrics are summarized. Finally, the application scenarios of AI in COPD research are elaborated, focusing on three aspects: early screening and diagnosis, disease monitoring and risk prediction, and disease classification and risk stratification. This paper summarizes the main research hotspots at home and abroad over the past five years with representative models, and analyzes and compares the advantages and limitations of each type of model in specific COPD tasks through comparative experiments. The study also outlines prospects for the future development of this field, aiming to provide theoretical references and insights for subsequent research.
4 Apr 2026
Read appraisal →International journal of nursing studies
The use of large language models in clinical documentation: A scoping review
BACKGROUND: Clinical documentation is essential for safe, high-quality care but has become increasingly complex, contributing to clinician burnout. Large language models offer potential to ease documentation by generating summaries, structuring data, and ensuring compliance. However, concerns remain regarding accuracy, bias, privacy, and regulatory risks. OBJECTIVE: To map current literature on large language models applications in clinical documentation, evaluating their benefits, limitations, and ethical considerations. INFORMATION SOURCES: Five electronic databases (i.e., PubMed, Scopus, CINAHL, Cochrane Library, and IEEE Xplore) covering peer-reviewed literature published in English between January 2009 and August 2025. METHODS: This scoping review followed Arksey and OMalleys framework and was reported in accordance with PRISMA-ScR guidelines. Screening, data extraction, and quality appraisal were conducted independently by multiple reviewers using Joanna Briggs Institute tools. Findings were synthesized using descriptive and narrative approaches. RESULTS: Forty-one studies met inclusion criteria, most originating from the United States. Large language models were primarily applied to clinical note generation, discharge summaries, and provider-patient encounter documentation. Key evaluation metrics included content accuracy, linguistic quality, and summarization performance. Large language models demonstrated potential to improve documentation efficiency and readability, with some studies reporting up to 40 % time savings. However, concerns about factual inaccuracies, hallucinations, and reduced performance in complex cases were common. Clinician perceptions were mixed. Some found notes generated by large language models helpful and well-structured, while others raised concerns about reliability, liability, and loss of clinical nuance. Ethical challenges included data privacy, security, and algorithmic bias, with varying levels of compliance across settings. CONCLUSIONS: Large language models hold significant promise for enhancing clinical documentation by improving efficiency, standardization, and clarity. However, their safe and effective use requires rigorous attention to accuracy, ethical safeguards, and clinician trust. Integration must support, rather than supplant, clinical reasoning and patient-centered care. Co-design with clinicians, real-world evaluation, and artificial intelligence literacy are essential to ensure that these technologies augment, not erode, professional judgment and care quality. REGISTRATION: Open Science Framework Registries (https://osf.io/m4h3q).
2 Apr 2026
Read appraisal →Clin Obstet Gynecol
Artificial Intelligence in Obstetrics: Current Applications, Opportunities, and Clinical Implementation Challenges
Artificial intelligence is transforming obstetric practice through applications in diagnostic imaging, risk prediction, and clinical decision-making. Deep learning algorithms have achieved diagnostic accuracy comparable to that of experienced clinicians. However, gaps persist between algorithmic capability and clinical implementation. Critical challenges include limited external validation and algorithmic bias. This review examines current AI applications in obstetrics across multiple clinical domains: automated fetal biometry, structural anomaly detection, prediction of pregnancy complications, and intrapartum fetal surveillance. It highlights persistent technical, ethical, and implementation barriers. Key recommendations include multicenter validation across diverse populations, explainable AI approaches, and creating strong regulatory frameworks.
1 Mar 2026
Read appraisal →Kidney360
Artificial Intelligence in Critical Care Nephrology: Current Applications, Emerging Techniques, and Challenges to Clinical Integration
Artificial intelligence (AI), including machine learning, deep learning, reinforcement learning, and generative AI, has the potential to advance critical care nephrology (CCN) by enhancing prediction accuracy, improving diagnostic capabilities, supporting clinical decision making, and streamlining workflow processes. Current applications in CCN include AKI prediction, nephrotoxin surveillance, intradialytic hypotension forecasting, and AI-guided dialysis and continuous KRT management, with performance often exceeding traditional models. However, the effect on patient-centered outcomes such as mortality, dialysis dependence, and cost-effectiveness remains uncertain. Emerging techniques, such as conformal prediction for calibrated risk estimates, causal inference for intervention modeling, and reinforcement learning for adaptive ultrafiltration, show promise in enhancing reliability, interpretability, and individualized care. Generative AI and large language models extend these applications to clinical documentation, reasoning, and patient education, while raising new challenges, including hallucinations, regulatory oversight, and clinician trust. Persistent barriers such as data heterogeneity, limited external validation, alert fatigue, and economic constraints hinder broad adoption. This review synthesizes the current evidence and outlines four priorities for advancing AI in CCN: ( 1 ) rigorous multicenter validation focused on clinical outcomes, ( 2 ) integration of uncertainty quantification and causal modeling into AI tools, ( 3 ) development of clinician-centered interfaces that minimize cognitive load, and ( 4 ) establishment of transparent, adaptive regulatory and governance frameworks. Realizing the promise of AI in CCN will require multidisciplinary collaboration, fairness and generalizability testing, and sustainable implementation strategies that align technologic innovation with measurable improvements in patient care.
1 Mar 2026
Read appraisal →BMC Palliat Care
Identification and mapping of mHealth interventions applied in palliative care in Sub Saharan Africa: a scoping review
BACKGROUND: Although palliative care is very essential for improving quality of life (QOL) by providing relief from pain and other distressing symptoms, its coverage in sub Saharan Africa remains remarkably low. Utilization of mHealth could be a fundamental healthcare system strengthening approach to enhance and promote palliative care in the region. There is paucity of evidence regarding evidence on the use of these interventions in SSA because it is a relatively new concept in the region. Therefore, there was need to identify and map the available/ applied mHealth interventions for palliative care in SSA basing on published and grey literature. AIM: To examine the nature and use of mHealth interventions for palliative care and factors influencing their use in sub Saharan Africa (SSA). METHODS: The review was based on the JBI methodology for scoping review. PUBMED, CINAHL, Embase and Google Scholar databases were searched for published articles with no time restrictions. Unpublished and grey literature was also searched for. Data was extracted from the retrieved sources and synthesized. The findings were reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) in a comprehensive and transparent way. RESULTS: A total of 224 articles were identified and 35 of these met the inclusion criterion and were considered for the review. The results show use of mHealth interventions for palliative care in 12 countries of the SSA. Only 12 (35.3%) were app-based, 9(26.5%) were SMS-text-based, 4(11.8%) were telemedicine-based,5(14.7%) were phone call-based, 3(8.8%) were mixed while 1(2.9%) used wearable-sensor-based mHealth interventions. Key uses of mHealth interventions in palliative care included health strengthening, psychosocial support to patients and families, enhancement of compliance to appointments for patients, promotion of adherence to medication, health education, monitoring, evaluation and research and capacity building for palliative care health professionals. A number of influencing factors were identified and they included wide-spread telephone ownership and access, great robustness of telephone and internet connectivity, high degree of perceived usefulness and ease of use, feasibility and acceptability of mHealth. Also, vital barriers such as low levels of technological advancement/expertise, irregularity and paucity of electric power supply, high likelihood to compromise patients' privacy and confidentiality and high illiteracy rates among the population were articulated by the articles. CONCLUSION: About a quarter of SSA countries have incorporated mHealth interventions for palliative care, underscoring the need for further research and development of the technologies. These approaches are quite beneficial to SSA, where there is a disproportionate disease burden and an inadequate healthcare system.
1 Feb 2026
Read appraisal →Ageing Res Rev
AI-driven aging digital twins: A roadmap for clinical translation in precision geriatrics
Aging and related diseases account for ⁓70 % of global mortality with declining physiological homeostasis and cognitive function increasing susceptibility to neurodegenerative disorders, cardiovascular diseases and cancer. Treatment is hampered by imprecise diagnostics and limited therapeutic options. Precision medicine involving artificial intelligence (AI)-powered Aging Digital Twin (ADT) technology which integrates multi-source heterogeneous data via advanced AI and deep learning has great potential. Virtual doppelgangers enable real-time health monitoring, predictive analytics and personalized interventions to facilitate early disease detection and optimize risk assessment and treatment strategies. Challenges remain and include age-stratified biases in training datasets and variations in data quality for older populations which may limit model generalizability. Critical concerns regarding data privacy, algorithmic transparency and clinical validation of ADT models remain unresolved. The current review examines the transformative potential of AI-enabled ADTs in precision medicine for aging and highlights technical and ethical limitations.
1 Jan 2026
Read appraisal →BMC Geriatr
Smart aging: integrating AI into elderly healthcare
Artificial Intelligence (AI) is transforming geriatric healthcare by improving early disease detection, optimizing patient management, and enhancing clinical decision-making. AI-driven tools, including machine learning (ML) algorithms, predictive analytics, and assistive robotics, are increasingly utilized to address challenges associated with aging, such as frailty, multimorbidity, polypharmacy, and fall prevention. These technologies can facilitate personalized medicine, improve diagnostic accuracy, and support healthcare administration by streamlining workflows and resource allocation. Despite its promising applications, the integration of AI in geriatric care presents several challenges, including issues related to data privacy, algorithmic bias, ethical considerations, and the digital literacy of older adults. Ensuring the inclusivity of AI models by incorporating diverse and representative datasets is essential to avoid disparities in healthcare delivery. Additionally, the role of AI must be carefully regulated to complement, rather than replace, human clinical expertise. This paper provides a comprehensive overview of AI applications in geriatric medicine, discussing its benefits, limitations, and future directions. By leveraging AI responsibly, healthcare systems can improve patient outcomes, reduce hospital readmissions, and promote aging in place. However, addressing existing challenges through interdisciplinary collaboration, policy development, and continued research is crucial to fully realizing AI's potential in elderly care.
1 Dec 2025
Read appraisal →Am J Obstet Gynecol
Can artificial intelligence improve the readability of patient education information in gynecology?
BACKGROUND: The American Medical Association recommends that patient information be written at a sixth-grade level to increase accessibility. However, most existing patient education materials exceed this threshold, posing challenges to patient comprehension. Artificial intelligence, particularly large language models, presents an opportunity to improve the readability of medical information. Despite the growing integration of artificial intelligence in healthcare, few studies have evaluated the effectiveness of large language models in generating or improving readability of existing patient education materials within gynecology. OBJECTIVE: To assess the readability and effectiveness of patient education materials generated by ChatGPT, Gemini, and CoPilot compared to American College of Obstetricians and Gynecologists and UpToDate.com. Additionally, to determine whether these large language models can successfully adjust the reading level to a sixth-grade standard. STUDY DESIGN: This cross-sectional study analyzed American College of Obstetricians and Gynecologists, UpToDate, and large language model-generated content, evaluating large language models for 2 tasks: 1) independent large language model-generated materials and 2) large language model-enhanced versions reducing existing patient information to sixth-grade level. All materials were assessed for basic textual analysis and readability using 8 readability formulas. Two board-certified obstetrician-gynecologists evaluated blinded patient education materials for accuracy, clarity, and comprehension. Analysis of variance was used to compare textual analysis and readability scores, with Tukey post-hoc tests identifying differences for both original and enhanced materials. An alpha threshold of P<.004 was used to account for multiple comparisons. RESULTS: Large language model-generated materials were significantly shorter (mean word count 407.9 vs 1132.0; P<.001) but had a higher proportion of difficult words (36.7% vs 27.4%; P<.001). American College of Obstetricians and Gynecologists and UpToDate materials averaged ninth-grade and 8.6-grade levels, respectively, while artificial intelligence-generated content reached a 10.6-grade level (P = .008). Although CoPilot and Gemini improved readability when prompted, no large language model successfully reached the sixth-grade benchmark, and ChatGPT increased reading difficulty. CONCLUSION: Large language models generated more concise patient education materials but often introduced more complex vocabulary, ultimately failing to meet recommended health literacy standards. Even when explicitly prompted, no large language model achieved the sixth-grade reading level required for optimal patient comprehension. Without proper oversight, artificial intelligence-generated patient education materials may create the illusion of simplicity while reducing true accessibility. Future efforts should focus on integrating health literacy safeguards into artificial intelligence models before clinical implementation.
1 Dec 2025
Read appraisal →Gerontologist
The AI-aging-enterprise: a political economy of aging and artificial intelligence
The current discourse on artificial intelligence (AI) in gerontology remains mostly on an interventionist level and focused on solving problems faced by individuals, leaving the wider social conditions that shape the relationship between aging and AI out of view. The considerable accumulation of power, particularly for technology development companies, in the development and implementation of AI, however, calls for a deeper and more complex analysis of the relationships between AI, older adults and the "aging enterprise." Building on the classical political economy of aging, and expanding it with concepts from material gerontology, we propose a "political economy of aging and AI" as a conceptual tool to analyze the relationships between (older) individuals, social and political structures, and (technological) materialities. We exemplify how such a political economy perspective enables a more complex consideration of the current discourse on (a) AI solutions, (b) AI ethics, and (c) AI policies. We close by summarizing our understanding of a political economy of aging and AI that centers materialities as the critical-gerontological focus of analysis and by outlining calls for action toward strengthening the power of older adults in the emerging AI-aging-enterprise.
1 Nov 2025
Read appraisal →Acad Pediatr
Caregiver and Pediatric Clinician Perspectives on Artificial Intelligence for Language Services
OBJECTIVE: Children and caregivers who use languages other than English (LOE) for pediatric health care experience inequities and poorer overall health. Advancement of large language models has raised questions about the future use of artificial intelligence (AI) for language access. This study explored the perspectives of caregivers who use LOE and pediatric clinicians on using AI to address unmet needs in health care translation and interpretation. METHODS: We conducted semistructured individual interviews with caregivers of pediatric patients who use LOE and pediatric clinicians about the use of AI language technologies in health care. A phone interpreter or fluent team member was used for LOE interviews. Recordings were transcribed, translated as needed, coded, and analyzed using inductive thematic analysis. RESULTS: Twenty caregivers using 11 different LOE and 22 clinicians participated. Themes were consistent across participants, though clinicians were more familiar with the concept of AI. Participants reported use of technologies (eg, Google Translate) for written communication and situations where verbal interpretation was perceived to be inadequate. They were concerned about AI accuracy, privacy, and loss of empathy, but hoped it could provide real-time document translation and more convenient verbal communication. Ease of use, validation, and equitable creation and access were critical for use. CONCLUSIONS: Caregivers and clinicians were open to using AI to fill gaps in translation and interpretation; however, robust validation of AI technology and complementary use with human interpreters and translators is needed. Future research, practice, and policy should focus on integrating AI while investing in human translation and interpretation.
1 Jan 2025
Read appraisal →