Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 26 appraisals
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 →Journal of robotic surgery
The need for a dedicated surgical consent framework for robotic telesurgery: a global call to action
Robotic telesurgery has progressed from experimental demonstrations to real-world clinical implementation across multiple continents. Landmark interconnected procedures, including intercontinental robotic prostatectomies, humanitarian implementations in underserved regions, and expansion of teleproctoring/tele-mentoring networks, showcase a rapidly maturing field. Despite this proliferation, no standardized, procedure-specific surgical consent for telesurgery exists in the medical literature. Current consent processes for robotic surgery, telemedicine, and surgical research fail to address critical elements unique to telesurgery, including latency thresholds, technical redundancies, network architecture, cybersecurity risks, remote surgeon responsibilities, and contingency planning. Our article aims to define the essential elements that should compose a dedicated telesurgery-specific informed consent and to highlight why current consent frameworks are insufficient for this rapidly expanding global practice. This expert-opinion paper synthesizes high-impact literature on consent ethics, digital surgery, telehealth governance, and robotic surgery; evaluates current consents used in pioneering telesurgery programs across North America, South America, Africa, Europe, and Asia; and incorporates multidisciplinary insights from surgery, anesthesiology, biomedical engineering, cybersecurity, telecommunications, ethics, and global health. We identify major gaps in contemporary consent frameworks, including (1) inadequate disclosure of connectivity-related risks; (2) insufficient explanation of data transmission, cybersecurity, and encryption; (3) absence of latency thresholds and safety-based interruption criteria; (4) lack of transparency regarding remote surgeon responsibility and handover protocols; and (5) limited patient education regarding contingency conversion and chain-of-care accountability. We thus propose a 9-component International Telesurgery Consent Framework, providing standardized language and structure for global implementation. The rapid expansion of telesurgery demands an equally rapid evolution in patient-centered informed consent. A dedicated telesurgery consent should become the ethical, medicolegal, and operational standard for any remote robotic clinical procedure. This paper serves as the basis for an international consensus effort involving global leaders in robotic surgery, telehealth, AI-enabled surgery, and digital health regulation.
14 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 →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 →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 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 →Journal of medical Internet research
Patients' Perspectives on the Implementation of AI in Radiological Diagnostics: Focus Group Study
BACKGROUND: Rapid developments in artificial intelligence (AI) will enable its widespread use in radiological diagnostics in the near future. Patients will then be confronted with findings generated with the help of AI. Understanding patients' perspectives on the use of this technology is one of the key factors for its successful implementation. OBJECTIVE: This qualitative study aimed to gain insight into patients' reasoning about the opportunities and risks of using AI in radiological diagnosis, to identify prerequisites for its acceptance, and to identify aspects that can promote trust in AI diagnoses, especially in scenarios of high personal concern. METHODS: A total of 7 focus groups were conducted with 34 patients (n=15, 44% female participants) aged between 23 and 85 years (mean 49.06, SD 17.08 y), recruited using purposive sampling strategies. Each focus group was audiotaped, transcribed, and analyzed using the method of structured qualitative content analysis. RESULTS: Study findings show that patients are open to the use of AI in radiological diagnostics. The basic prerequisites for this are (1) scientific evidence of safe outcomes that are more accurate and faster than those without AI; (2) recognizable added value in patient care; (3) transparency in the use of AI and disclosure to the patient; (4) comprehensive, binding measures for quality assurance; and (5) the use of AI solely to support the physician. However, the results indicate that further criteria are important for patients to be willing to choose a radiologist who uses AI and to trust AI diagnoses. In situations where they are personally affected, patients fear that physicians will place too much trust in the AI result and that the physician-patient relationship will become dehumanized. Therefore, the physicians' abilities and functions that inspire trust from the patients' perspective must come into play. These include (1) an independent diagnosis by the physician that takes into account not only the clinical context but also the individuality of the patient, (2) a comprehensible explanation of the pros and cons of using AI for patients and clear communication of AI output, and (3) a humane and empathetic physician-patient relationship, which shows that the physician continues to feel responsible for the patient. CONCLUSIONS: The results of the study underscore that a high quality of the entire "sociotechnical" system is an essential prerequisite for patient acceptance of the use of AI in radiological diagnostics and for trust in AI diagnoses. The further development of AI performance must go hand in hand with the creation of framework conditions for its use that meet patients' expectations of the role of the physician and ensure a trust-building physician-patient relationship. The study provides valuable insights into how such integration of AI into radiological practice can be achieved.
26 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
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 →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 →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 →The Journal of clinical ethics
Ethical Responsibility in the Off-Label Use of AI in Medical Imaging
AbstractArtificial intelligence in medical imaging (AI-MI), categorized by the U.S. Food and Drug Administration (FDA) as "Software as a Medical Device," can offer significant benefit to medical care. FDA approval-represented by a "label"-indicates that the device has demonstrated safety and efficacy for a specific clinical task and population for which the device is intended. Off-label use of AI-MI outside of the task and/or population can be medically and ethically problematic, yet benefits may be reduced or missed if entirely prohibited. Ethical responsibility frameworks can assist in evaluating and establishing boundaries and opportunities in off-label use of AI-MI. In this work, we review three different models for ethical responsibility and apply them to off-label use of AI-MI. We find that a future-looking collective responsibility framework, involving stakeholders such as industry developers, healthcare organizations, clinicians, and, most importantly, patients and/or their representatives, promotes proactive and practical actions to help ensure safe and ethical use of AI-MI off-label. By adopting this approach, stakeholders can balance innovation with safeguards, advancing patient-centered care while mitigating potential harms associated with off-label AI-MI use. This framework is crucial as AI-MI becomes increasingly integral to clinical decision-making processes.
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 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 →Palliative & supportive care
Relational-temporal dignity at the end of life: Ethical foundations for the DiRePal model in the algorithmic age
OBJECTIVES: The increasing presence of artificial intelligence (AI), electronic patient-reported outcomes (ePROMs), and digital infrastructures in palliative care is transforming how clinical encounters are organized and how suffering is interpreted. These technological shifts heighten the risk of relational compression and a reduction of dignity to measurable outputs. This paper proposes the DiRePal model (Relational-Temporal Dignity in Palliative Care) as a philosophical framework to re-examine dignity beyond coherent narrative identity or autonomy-centered ethics, emphasizing relational presence, temporal sensitivity, and structural conditions of care. METHODS: A philosophical-ethical analysis informed by narrative identity (P. Ricoeur), ethics of alterity (E. Levinas), capabilities theory (M. Nussbaum), and care ethics (J. Tronto). Critical readings of dignity frameworks, AI ethics, and digital health literature were synthesized to develop a relational-temporal account of dignity and 2 operational concepts: the temporal dignity indicator and the architecture of prudence. RESULTS: While digital tools can enhance communication and support anticipatory care, they also risk reducing patients to data profiles, narrowing listening practices, and eroding opportunities for narrative, silence, and relational presence. The DiRePal model reframes dignity as a fluctuating, co-constructed achievement that depends on temporal attentiveness, ethical listening, institutional conditions, and prudent integration of AI and ePROMs. It further expands dignity to include post-biographical dimensions such as memory, grievability, and digital legacy. SIGNIFICANCE OF RESULTS: End-of-life care in the algorithmic age requires an ethics that recognizes dignity as relational, temporal, and structurally mediated. The DiRePal model offers clinicians and institutions a conceptual grammar to resist technological reductionism, protect time for presence, and safeguard the narrative and post-biographical continuity of persons whose voices may be fragmented, vulnerable, or digitally extended.
11 May 2026
Read appraisal →Palliative & supportive care
Definitional confusion in nurses' and physicians' attitudes toward euthanasia in India: A systematic review and meta-analysis
INTRODUCTION: In India, the term euthanasia is frequently used to describe a wide range of end-of-life practices, including withdrawal of life-sustaining treatment and administration of lethal drugs. Such usage diverges from more narrow definitions that restrict euthanasia to the intentional administration of lethal drugs. OBJECTIVES AND SIGNIFICANCE OF THE RESULTS: This systematic review and meta-analysis examines how euthanasia has been defined and operationalized in quantitative studies of Indian physicians' and nurses' attitudes, and estimates the prevalence of approval when euthanasia is defined narrowly. METHODS: Following PRISMA 2020 guidelines, searches were conducted in PubMed, EMBASE, PsycINFO, and CINAHL for studies published from 2010 onward. Two investigators independently screened studies, extracted data, and assessed risk of bias using the Mixed Methods Appraisal Tool (MMAT). Studies were included in the meta-analysis only if they reported attitudes toward euthanasia as narrowly defined. RESULTS: Nine studies met inclusion criteria for the systematic review. Definitions of euthanasia varied considerably, and several studies combined attitudes toward treatment withdrawal with attitudes toward the administration of lethal drugs. Four studies reporting on 519 nurses and physicians provided data suitable for meta-analysis. Approval of euthanasia, defined as the intentional administration of lethal drugs, ranged from 12% to 20%, with a pooled prevalence of 16% (95% confidence interval: 0.13-0.19). This prevalence is notably lower than in earlier reports on Indian healthcare professionals' attitudes. CONCLUSION: Definitional inconsistency substantially affects reported attitudes toward euthanasia in Indian research. When euthanasia is defined narrowly, approval among healthcare professionals is low and consistent across studies. These findings highlight the need for conceptual clarity in future research to support accurate interpretation of empirical data and to strengthen the contribution of studies on ethical attitudes to ethical and palliative care scholarship.
10 May 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 →JMIR formative research
Global Mapping of Telemedicine Regulation and Ethical Safeguards: Mixed Methods Exploratory Document Analysis
BACKGROUND: Telemedicine has become central to digital health strategies, yet the regulatory environment that shapes ethical safeguards and equitable access remains uneven and incompletely assessed across countries. Legal and normative instruments matter because they define requirements for privacy, consent, accountability, professional readiness, and barrier reduction. OBJECTIVE: This study aimed to map the current global landscape of normative instruments related to telemedicine and identify which ethical and social safeguards are explicitly addressed, with particular attention to equity. METHODS: We conducted a document analysis guided by the READ (ready the materials, extract data, analyze data, and distill findings) framework. From February 2024 to February 2026, we conducted a structured web-based search across all World Health Organization (WHO) member states with no language restrictions, using official government sources, statutory professional regulators, and institutional publication channels. Retrieval combined internal site searches, direct navigation, external search engine queries, and targeted snowball sampling to identify currently in-force instruments. Two researchers independently extracted and coded data using a predefined codebook. We operationalized 10 binary items covering regulatory presence and scope (questions 1 and 2), safeguards for data protection (question 3), consent and disclosure (questions 4 and 5), prior in-person prerequisites (question 6), monitoring (question 7), training requirements (question 8), and equity (questions 9 and 10). We summarized frequencies overall and stratified by WHO region and World Bank income group and conducted a qualitative thematic analysis of included normative instruments. RESULTS: Of the 194 WHO member states, 81 (41.8%) had at least one current normative instrument related to telemedicine in force. Among these, 72.8% (59/81) defined telemedicine or telehealth. Data protection provisions were most common (73/81, 90.1%), followed by mandatory informed consent (n=71, 87.7%) and monitoring mechanisms (n=65, 80.2%). Fewer countries required disclosure of telemedicine limitations (n=36, 44.4%) or mandated telemedicine-specific training (n=26, 32.1%). Prior in-person consultation requirements were uncommon (n=8, 9.9%). Equity-related safeguards were uneven: 51.9% (n=42) referenced justice, equity, or nondiscrimination, whereas 30.9% (n=25) included concrete barrier reduction provisions (eg, digital inclusion or accommodations for people with disabilities and minors). CONCLUSIONS: Telemedicine regulation is becoming more common, but both coverage and safeguarding content remain uneven. While privacy, consent, and monitoring are frequently addressed where regulation exists, disclosure, physician competency, and operational equity measures are less consistently specified. Strengthening telemedicine governance will require translating ethical commitments into enforceable standards that address digital determinants of access and protect groups at risk of exclusion.
25 Apr 2026
Read appraisal →BMJ open
From legalisation to practice: implementation of newly legalised health interventions in healthcare settings - a systematic review.
OBJECTIVES: Over the last decade, a growing number of health interventions (eg, medical assistance in dying and mitochondrial donation) have become legalised or decriminalised globally. Newly legalised health interventions share characteristics that are distinct from other health interventions, making their implementation more challenging. They are often highly emotive, controversial and associated with strong opinions and ethical dilemmas, with some of them being high-stake and irreversible. This study aimed to identify, systematise and map the factors that affect the implementation of health interventions that have recently been legalised. DESIGN: A systematically conducted review. DATA SOURCES: PubMed, Scopus, EMBASE and CINAHL were searched to identify studies published between 2014 and 2024. ELIGIBILITY CRITERIA FOR SELECTING STUDIES: We included studies if they evaluated the implementation of health interventions that were newly legalised or newly decriminalised. DATA EXTRACTION AND SYNTHESIS: Data were extracted and synthesised through descriptive analysis. Both deductive and inductive thematic analyses were applied to map the barriers, facilitators and implementing strategies that influence the implementation of newly legalised health interventions in healthcare settings. RESULTS: The search strategy yielded 1510 publications, of which 78 were included in this review. Findings showed that several newly legalised health interventions, including medical assistance in dying (n=56 studies); medical abortion (n=13); assisted human reproduction (n=3); psychedelic-assisted therapies (n=3); use of medical cannabis (n=2) and use of biosimilars (n=1) were addressed. The analysis identified a total of 880 diverse barriers, facilitators and strategies in five domains across system, organisational and individual levels: (1) patients/service users/consumers; (2) healthcare providers; (3) healthcare organisation; (4) legal processes and (5) system. These were further divided into 27 themes of barriers, 18 themes of facilitators and 17 themes of strategies. CONCLUSIONS: Implementing newly legalised health interventions is complex. Our findings can support the development of an implementation plan for the spread and scaling of future health interventions, maximising the impact of interventions and making them accessible to more people and health organisations.
24 Apr 2026
Read appraisal →Medical education online
Medical students' attitudes towards older persons - a systematic review and meta-analysis
OBJECTIVE: Ageism negatively affects older people's health, well-being, and quality of care. Identifying ageism and attitudes toward older adults may help identify medical students motivated to work with a growing older population. We conducted a systematic review of interventions to reduce ageism among medical students. METHODS: We conducted a systematic literature retrieval in Medline, the Cochrane Library, Epistemonikos and PubMed via Ovid, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses, from database inception to August 23rd, 2024. We searched for studies examining medical students' attitudes towards older persons, including interventions to influence these attitudes. RESULTS: The systematic review included 21 studies. Study samples ranged between 29 and 415, with a pooled total of 3940 medical students, the majority conducted in the USA. The meta-analysis of students' attitudes before and after all interventions included a total of 2004 students, yielding a non-significant standardized mean difference (SMD) of 0, 12 (95% CI -0, 01 to 0, 26). The meta-analysis of students' attitudes before and after all interventions (35-37, 39, 41, 45, 48, 50), after removing UCLA-GAS and R-GAS, included a total of 1344 students with a standardized mean difference of 0, 23 (95% CI 0, 06 to 0, 40). Analysis of empathy-based teaching interventions included 1652 students, and showed a statistically significant effect, SMD 0, 18 (95% CI 0, 01 to 0, 34). Analysis of 352 students receiving traditional teaching methods yielded no effect on ageism, with an SMD -0, 13 (95% CI -0, 81 to 0, 54). High heterogeneity (I2 75-85%) affects our confidence in the effect estimates. CONCLUSION: Empathy-based interventions seem to improve medical students' attitudes towards older persons. Traditional teaching methods may increase negative attitudes. Considering the high heterogeneity, the results should be interpreted cautiously. Adding empathy-based components to medical curricula could combat ageism. More studies are needed to examine whether the results of studies from North America are valid globally.
23 Apr 2026
Read appraisal →JMIR formative research
Public Perceptions of AI in Medicine and Implications for Future Medical Education: Cross-Sectional Survey
BACKGROUND: The integration of artificial intelligence (AI) into clinical practice is contingent on public trust. This trust often depends on physician oversight, yet a significant gap exists between the need for AI-competent physicians and the current state of medical education. While the perspectives of students and experts on this gap are known, the views of the US general public remain largely unquantified. OBJECTIVE: This study aimed to assess US public perceptions regarding AI in medicine and the corresponding emergent needs for medical education. We specifically sought to quantify public trust in different diagnostic scenarios, concerns about physician overreliance on AI, support for mandatory AI education, and priorities for the future focus of medical training. METHODS: We conducted a cross-sectional, web-based survey of adults in the United States in November 2025. Participants (N=524) were recruited via SurveyMonkey Audience. We calculated descriptive statistics, frequencies, proportions (percentages), and 95% CIs for all main survey items. RESULTS: A total of 524 participants completed the survey. Most (n=329, 62.8%; 95% CI 58.6%-66.9%) placed the most trust in a physician's diagnosis based on their expertise alone; only 7.8% (n=41; 95% CI 5.5%-10.1%) trusted an AI-first diagnostic model. Trust was highly contingent on training: 93.9% (n=492) of participants rated formal physician training on AI limitations as "essential" or "very important." Widespread concern about physician overreliance on AI was reported, with 81.1% (n=425) being "very concerned" or "extremely concerned." Consequently, 85.1% (n=446) agreed or strongly agreed that training on AI use, ethics, and limitations should be mandatory in medical school. When asked about future educational priorities, 70.2% (n=368; 95% CI 66.3%-74.1%) believed that medical education should focus on human-centered skills (eg, empathy and communication) over clinical skills. CONCLUSIONS: The US public expressed conditional trust in medical AI, strongly preferring physician-led and critically supervised models. These findings reveal a clear public mandate for medical education reform. The public expects future physicians to be mandatorily trained to appraise AI, understand its limitations, and refocus their professional development on the human-centered skills that technology cannot replace.
18 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 →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 →Aesthetic Plast Surg
The Rise of Intelligent Plastic Surgery: A 10-Year Bibliometric Journey Through AI Applications, Challenges, and Transformative Potential
BACKGROUND: Driven by advancements in deep learning, surgical robots, and predictive modeling technologies, the integration of artificial intelligence (AI) and plastic surgery has expanded rapidly. Although AI shows the potential to enhance precision and efficiency, its clinical integration faces challenges, including ethical concerns and interdisciplinary complexity, which require a systematic analysis of research trends. METHODS: The CiteSpace and VOSviewer software were used to conduct a quantitative analysis of 235 documents in the core collection of Web of Science from 2016 to 2024. Co-citation networks, keyword co-occurrence, burst detection, and cluster analysis were employed to map the research trajectories. The inclusion criteria gave priority to studies that explicitly incorporated artificial intelligence into surgical designs or outcomes. The contributions of countries, institutions, and authors were evaluated through centrality indicators. RESULT: Publications related to artificial intelligence have grown exponentially, with the USA, Germany, and Canada leading research output. Harvard and Stanford Universities dominate in terms of institutional contributions, but cross-institutional collaboration remains limited. The keyword cluster highlights the innovations of artificial intelligence in breast reconstruction, facial analysis, and automated grading systems. Burst terms such as "deep learning," "risk assessment," and "attractiveness" underscore AI's role in optimizing surgical outcomes, but they also expose biases against Western-centric beauty standards. Ethical concerns, dataset diversity gaps, and overreliance on AI-driven decisions have become key obstacles. CONCLUSION: The integration of artificial intelligence in plastic surgery goes beyond the utility based on tools and into data-informed surgical engineering. The persistent gap in collaboration and dataset diversity highlights the need for global, interdisciplinary efforts to address technical and ethical challenges while advancing AI's clinical utility. Future research must prioritize transparency, inclusivity, and collaborative innovation to realize AI's transformative potential while mitigating risks. LEVEL OF EVIDENCE IV: This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .
1 Mar 2026
Read appraisal →