Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 44 appraisals
Minimally invasive therapy & allied technologies : MITAT : official journal of the Society for Minimally Invasive Therapy
Ventral hernia repair in emergency settings. A machine learning model to predict post-operative complications.
BACKGROUND: Emergency ventral hernia repair remains a challenging procedure due to patient instability, contaminated surgical fields, and heterogeneity in hernia types and operative techniques. Predicting postoperative complications in this setting is difficult using traditional statistical methods. Machine learning (ML) may offer improved predictive accuracy by recognizing nonlinear patterns among multiple perioperative factors. METHODS: A retrospective multicenter analysis was performed using data from the ACTIVE (Acute Treatment for Incisional Ventral Hernias) study, including 557 adult patients undergoing emergent ventral hernia repair between 2018 and 2021 in 31 Italian surgical centers. Demographic, preoperative, intraoperative, and postoperative variables were analyzed. Three ML algorithms-Decision Tree, Random Forest, and Deep Learning Neural Network-were trained and validated using five-fold cross-validation after class balancing with SMOTE. Model performance was compared with traditional logistic regression using accuracy, area under the ROC curve (AUC), and F1 score. RESULTS: Postoperative complications occurred in 181 patients (32.5%), while major complications (Clavien-Dindo ≥ II) occurred in 10%. Random Forest achieved the best performance (AUC 0.95, accuracy 0.88, F1 score 0.86), outperforming logistic regression (AUC 0.82, accuracy 0.78). The most influential predictors were operative duration, ASA score, and sepsis for overall complications, while bowel obstruction and BMI were key factors for major complications. Surgical approach (open vs. laparoscopic) did not independently correlate with adverse outcomes, highlighting the complexity of patient- and case-specific interactions. CONCLUSIONS: Machine learning models can accurately predict postoperative complications following emergent ventral hernia repair, surpassing traditional regression methods. These findings suggest that ML-based decision tools could support risk stratification and optimize surgical planning in high-risk emergency settings. Prospective validation is warranted to integrate AI-assisted prediction into perioperative clinical workflows.
3 Aug 2026
Read appraisal →Journal of the American Medical Informatics Association : JAMIA
Sociodemographic bias in large language model clinical trial screening
OBJECTIVE: To assess whether large language model (LLM)-based clinical trial screening judgments vary by patient sociodemographic characteristics. MATERIALS AND METHODS: We conducted a cross-sectional evaluation of Phase II-III US adult randomized controlled trial (RCT) protocols (2023-2024). Physician-validated clinical vignettes were evaluated in a control version and 33 sociodemographic identity variants differing only by labels. Nine LLMs assessed eligibility and related domains. Mixed-effects models estimated adjusted differences vs control. RESULTS: Across 58 protocols and 5.3 million evaluations, eligibility judgments were largely stable across identities. Race and ethnicity showed minimal effects after accounting for socioeconomic status. Homelessness produced the largest negative eligibility shift and pronounced effects in adherence, resources, and trust. DISCUSSION AND CONCLUSION: LLMs applied explicit eligibility criteria consistently, but disparities emerged in domains requiring inference about behavior or resources, underscoring the need for careful deployment to promote fair trial access.
2 Aug 2026
Read appraisal →Acta odontologica Scandinavica
Predicting gingival embrasure risk after invisible orthodontics using multimodal data and machine learning
OBJECTIVE: To develop and validate a risk prediction model for gingival embrasures after clear aligner therapy using multimodal oral data. METHODS: A retrospective study of 340 patients (December 2022-June 2025) was randomly divided into training (n = 238) and validation (n = 102) sets (7:3). Univariate analysis, multivariate logistic regression, and least absolute shrinkage and selection operator regression were applied to identify independent risk factors. Three machine learning models - random forest (RF), logistic regression, and support vector machine - were constructed based on seven core variables. Model performance was assessed using area under the receiver operating characteristic curve (AUC), calibration curves, decision curve analysis, and Shapley Additive Explanations (SHAP) values for interpretability. RESULTS: No significant baseline differences existed between sets (p > 0.05). Seven indicators were identified (p < 0.05). Multivariate analysis confirmed percentage of bleeding on probing-positive sites, interproximal alveolar bone height, and relative movement of adjacent teeth at target site as independent risk factors, while gingival thickness, proximal contact area, interdental papilla height, and buccal/lingual bone plate thickness were protective factors (p < 0.05). The RF model performed best: training AUC = 0.849 (95% CI: 0.786-0.912), validation AUC = 0.815 (95% CI: 0.720-0.910), with good calibration and net benefit. SHAP analysis highlighted gingival thickness and interproximal alveolar bone height as key predictors. CONCLUSION: A risk prediction model for post-clear aligner gingival embrasures was successfully developed and validated using multimodal oral data, with RF as the optimal algorithm. The model exhibits good discrimination, calibration, and clinical utility, which can be used as an objective auxiliary tool for individualized risk prediction following clear aligner therapy and supplement traditional clinical empirical judgment.
23 July 2026
Read appraisal →International ophthalmology
The relationship between diabetic retinopathy and intestinal microbiota: a systematic review and meta-analysis
PURPOSE: The results of human observational studies on the correlation between gut microbiota and diabetic retinopathy (DR) are discrepant. This meta-analysis aimed to evaluate the specificity of the gut microbiota in DR patients compared to patients with type 2 diabetes mellitus (T2DM). METHODS: All published literature up to October 2024 was searched by two researchers on PubMed, Embase and Web of Science databases. Diversity and gut microbiota composition were the main outcomes. The meta-analysis was conducted in Review Manager (RevMan) Version 5.3. RESULTS: Eight studies, investigating gut microbiota by collecting stool samples, conducted in China and India were included in this meta-analysis, involving a total of 486 individuals in the T2DM (n = 258) and DR (n = 228) groups. No significant difference in alpha-diversity was observed between T2DM and DR patients. The pooled estimate showed that, at the phylum level, the abundances of Patescibacteria, and Synergistetes were significantly lower, and Verrucomicrobia were considerably higher in DR patients than in T2DM patients. At the genus level, DR patients had an increase in Bacteroides compared to T2DM. CONCLUSIONS: In this meta-analysis with a small number of studies and relatively high heterogeneity, changes in gut microbiota were associated with DR, commonly reflected by a reduction in beneficial species and an increase in pathogenic species influencing metabolic pathways.
10 July 2026
Read appraisal →BMC medical research methodology
Variable selection for clinical prediction models in low-dimensional data - a simulation study comparing traditional regression and machine learning methods
PURPOSE: A wide range of methods exist for developing a clinical prediction model (CPM) and for performing variable selection. Our purpose was to develop a fair simulation study design and to investigate the properties, strengths, and weaknesses of different methods to predict a continuous outcome in low-dimensional data situations. METHODS: In this simulation study, we conducted a neutral comparison of traditional (linear regression with stepwise selection) and machine learning (regularized regression with elastic net, gradient boosting, random forest) variable selection strategies to derive a CPM. The generated datasets included a total of 15 variables, with 8 of those being predictor variables. Four data- and outcome-generating mechanisms with increasing complexity produced data structures typical for biomedicine covering linear associations and gradually introducing non-linear and non-additive elements into the data structure. RESULTS: All methods generally performed better with increasing sample size and less noise in the data. Gradient boosting with regression models and with trees as base learners, and the elastic net regularized regression included nearly all variables (i.e., both the predictor and non-predictor variables), especially with increasing sample size. The linear regression model with stepwise selection (LMSS) showed the best trade-off between correctly including the predictors and excluding the non-predictor variables in most of the scenarios, even when the functional form of continuous predictors deviated from linearity. In more complex data, variable selection using the Boruta or Hapfelmeier approach for random forest performed similar to LMSS. CONCLUSION: The sample size must be sufficiently large to enable the methods to reliably identify the predictor variables and to ensure that the developed CPMs are accurate and well-calibrated. LMSS revealed good properties and the random forest with the Boruta or Hapfelmeier approach are suitable alternatives if complex associations between predictors and outcomes are assumed.
9 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 →Cancer causes & control : CCC
Post-diagnosis physical activity in relation to mortality among gynecological cancer survivors
PURPOSE: Physical activity may play a supportive role in cancer survivorship. However, evidence on the association between post-diagnosis physical activity and mortality among women with gynecological cancer remains limited and inconsistent. METHODS: We conducted a systematic review of the literature published between 1949 and January 2026. Eligible observational studies were identified, and random-effects meta-analyses were performed to estimate pooled hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between post-diagnosis physical activity and all-cause mortality among women diagnosed with gynecological cancer. RESULTS: A total of ten eligible studies on endometrial, ovarian, and cervical cancer were included, collectively reporting 3,867 deaths. High levels of post-diagnosis physical activity, compared with low levels, were associated with lower mortality (HR: 0.65; 95% CI 0.54-0.78). This inverse relationship was evident in both endometrial and ovarian cancer survivors (endometrial cancer: HR: 0.60; 95% CI 0.43-0.83; ovarian cancer: HR: 0.71; 95% CI 0.58-0.86). Medium levels of physical activity tended to be inversely associated with mortality (HR: 0.88; 95% CI 0.76-1.02). CONCLUSION: Higher levels of physical activity after a gynecological cancer diagnosis were associated with improved survival. The results suggest that physical activity may represent a modifiable lifestyle factor with the potential to improve long-term outcomes among gynecological cancer survivors. IMPLICATIONS FOR CANCER SURVIVORS: This supports the potential value of integrating physical activity into survivorship care, although further high-quality prospective studies are needed to strengthen causal inference.
6 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 →Language, speech, and hearing services in schools
Enhancing Clinical Practice Through Research: Tools for School-Based Speech-Language Pathologists
PURPOSE: Speech-language pathologists (SLPs) in the school setting need to efficiently locate, evaluate, and utilize research evidence in their clinical practice. This tutorial provides resources focused on the foundational aspects of evidence-based practice (EBP) and its application for SLPs to implement research into clinical practice. METHOD: This tutorial (a) reviews study design and discusses clinical questions best suited to each; (b) provides key tips for extracting information from main sections of research articles; (c) discusses the use of generative artificial intelligence (AI) as a clinical and research tool, including actionable steps and examples for using AI in EBP; and (d) provides detailed strategies to increase interaction with research beyond reading articles. RESULTS: School-based SLPs can utilize the information and strategies provided to increase their confidence in evaluating research literature, applying information, and incorporating resources from research articles into their clinical practice. CONCLUSIONS: This tutorial equips school-based SLPs with actionable steps and concrete tools to increase ease and efficiency of applying research evidence into clinical practice. It provides guidance on different study designs, practical strategies to review research articles, and a case study demonstrating these strategies as well as discusses the potential use of AI to support clinicians in generating clinical questions and the steps of the EBP process. SUPPLEMENTAL MATERIAL: https://doi.org/10.23641/asha.31652170.
4 July 2026
Read appraisal →Experimental eye research
Artificial intelligence applications in OCT and OCTA for diabetic retinopathy: A systematic review
PURPOSE: Diabetic retinopathy (DR) is a leading cause of vision impairment worldwide. Optical coherence tomography (OCT) and OCT angiography (OCTA) provide detailed retinal imaging, enabling early detection of microvascular changes. This study aims to systematically review artificial intelligence (AI), particularly deep learning (DL), applications for DR detection and analysis using OCT and OCTA images. METHODS: A comprehensive literature search was conducted across PubMed, Web of Science, Scopus, IEEE Xplore, and Embase for studies published up to March 2026. A total of 1007 articles were identified, of which 175 studies met the inclusion criteria following the PRISMA study selection process. RESULTS: DL-based approaches consistently demonstrated superior performance compared to traditional machine learning (ML) methods, with reported AUC values typically ranging from 0.90 to 0.99 across classification and segmentation tasks. Convolutional neural networks (CNNs), Vision Transformers (ViTs), and encoder-decoder architectures such as U-Net showed strong performance in detecting key DR biomarkers, including microaneurysms, macular edema, and neovascularization. However, performance variability was observed depending on dataset size, imaging modality, and annotation quality. CONCLUSIONS: AI-driven analysis of OCT and OCTA images offers significant potential for automated DR detection. Despite promising results, challenges such as limited public datasets, lack of cross-institutional validation, and model interpretability remain. Future research should focus on multimodal integration, explainable AI, and large-scale validation to enhance clinical applicability.
2 July 2026
Read appraisal →International journal of gynaecology and obstetrics: the official organ of the International Federation of Gynaecology and Obstetrics
Integrated artificial intelligence and omics for prediction and monitoring of pre-eclampsia
Pre-eclampsia is a difficult pregnancy condition that causes high blood pressure and can lead to health complications in both mother and newborn, resulting in a higher fatality rate. It presents with a wide range of symptoms and lacks specific indicators, as the contemporary diagnostic techniques, including proteinuria testing and blood pressure measurements, are not reliable. The current evolution in artificial intelligence (AI) technology tends to show a promising transformation of pre-eclampsia management. AI algorithms are applied to process larger sets of clinical, biochemical, and image data that facilitate timely medical interventions by bringing up the early-onset and severity of pre-eclampsia. By analyzing the red cell distribution width (blood test indicators for pre-eclampsia), it is recognized as a cost-effective way of detecting inflammation. The application of AI technology on non-invasive diagnostic (wearable) devices enables continuous monitoring with imaging techniques for the placenta and retina via cloud-based systems. These developments are not only applied for early detection of pre-eclampsia, but also assist decision making capabilities in both high- and low-resource environments. This article explains how the growing use of AI is changing the way that pre-eclampsia is understood and managed, with the aim of improving accuracy and offering more personalized care for pregnant women.
1 July 2026
Read appraisal →The Cochrane database of systematic reviews
Prognostic models for predicting intensive care unit admission or mortality in critically ill adults not yet been admitted to the intensive care unit
OBJECTIVES: This is a protocol for a Cochrane Review (prognosis). The objectives are as follows: To determine whether prognostic models can be used for predicting the occurrence of ICU admission and mortality within the index hospital admission, and post-discharge survival at reported follow-up time points, up to one year, in critically ill adults not yet admitted to the ICU. The objective in PICOTS format is as follows. POPULATION: critically ill adults not yet admitted to the ICU Index prognostic model: available prognostic models with or without external validation Comparator: not applicable Outcomes: ICU admission, ICU and hospital mortality, and post-discharge survival Timing: for ICU admission or transfer, ICU mortality, and hospital mortality outcomes, the timing will be in hospital. For post-discharge survival, the timing will be up to 28 days, 1 to 3 months, 6 months, and 1 year after hospital discharge. SETTING: acute hospital wards and emergency care departments.
30 June 2026
Read appraisal →European journal of pediatrics
Feasibility of image-based 3D avatar anthropometry assessment of pediatric obesity: a pilot study.
UNLABELLED: The purpose of this study is to evaluate the feasibility and validity of 3D avatar-based anthropometry for assessing anthropometric indicators of overweight and obesity in children based on artificial intelligence-derived 3D body reconstruction. This cross-sectional study included 171 children aged 8-10 years from five primary schools in southern Spain. Due to technical constraints of the reconstruction software, 75 children meeting predefined image quality and body weight requirements (≥ 30 kg) were eligible for 3D avatar analysis. Manual anthropometric measurements were obtained following International Society for the Advancement of Kinanthropometry (ISAK) standards and compared with avatar-derived circumferences. Agreement was assessed by using Wilcoxon tests, Spearman correlations, Bland-Altman analyses, and false discovery rate-adjusted comparisons. Avatar-derived circumferences were consistently greater than manual measurements, although waist-to-hip and waist-to-stature ratios demonstrated no significant differences. Avatar and manual circumferences were moderately to strongly correlated (Spearman's ρ = 0.62-0.72; adjusted p < 0.05). However, the mean absolute errors ranged from 6.21 to 7.11 cm, and Bland-Altman analysis revealed systematic overestimation with wide limits of agreement, particularly for waist and hip circumferences. Despite these discrepancies, both methods similarly discriminated between weight status categories. CONCLUSION: Image-based 3D avatar anthropometry is a feasible, noninvasive approach for population-level screening and research in pediatric settings. Although the system reliably captures relative differences and body proportions, its current accuracy is insufficient for individual-level clinical assessment. Further algorithmic refinement and validation in lighter and more diverse pediatric populations are needed before clinical implementation. WHAT IS KNOWN: • Childhood obesity remains a major public health challenge requiring accurate monitoring. • Anthropometric assessment still relies mainly on manual measurements, while AI-based 3D body modeling offers a promising non-invasive alternative. WHAT IS NEW: • Image-based 3D avatar anthropometry is feasible in school settings but is constrained by technical requirements. • Avatar measurements are correlated with manual anthropometry and preserve weight status discrimination but demonstrate systematic overestimation and limited agreement.
29 June 2026
Read appraisal →Science advances
AI-driven tripartite classification for optimizing wearable bioelectronics in depression management
Current disease-sensing devices primarily focus on distinguishing between healthy and diseased states, effective for diagnosis but limited in guiding optimal intervention timing for prevention. We developed a tripartite framework identifying pre-disease state in depression, a reversible phase preceding irreversible onset. Using complex systems theory, we analyzed early-warning signals emerging as biological systems approach critical transitions. Continuous monitoring of nine multimodal biomarkers-spanning electrophysiological, behavioral, and biological-enabled classification into normal, pre-disease, and disease states by quantitatively defining critical points. An artificial intelligence agent classified disease states with 95.2% accuracy using multimodal data, enabled by ultrasoft neural probes for stable, low-damage recordings. Therapeutic validation with a skin-attachable wireless vagus nerve stimulator integrating soft three-dimensional electrodes demonstrated superior efficacy during pre-disease states. Subjects treated during pre-disease showed faster recovery and greater therapeutic responses, while those treated after disease onset failed to achieve full recovery. This framework provides evidence-based rationale for early intervention.
27 June 2026
Read appraisal →International journal of hyperthermia : the official journal of European Society for Hyperthermic Oncology, North American Hyperthermia Group
The role of capacitive hyperthermia as an adjunct treatment in oncology: a systematic review of randomized phase III trials.
INTRODUCTION: Capacitive hyperthermia (cHT) has been established for superficial tumors but the efficacy of cHT for intermediate or deep-seated tumors remains a subject of debate. This review is an evaluation of the current clinical evidence on the effectiveness of cHT in the treatment of tumors that are located deeper within the body. METHODS: A systematic review was conducted in accordance with the PRISMA guidelines of Phase III randomized controlled trials that included cancer patients of all ages and sexes with confirmed malignant tumors undergoing radiotherapy (RT) and/or chemotherapy (CT), with or without additional cHT. The literature search was conducted in PubMed and the Cochrane Register of Controlled Trials (CENTRAL). The primary endpoints included overall survival (OS), tumor response (TR; encompassing complete response (CR)), local control (LC), and cHT treatment-related toxicity (TRT). RESULTS: A total of nine randomized phase III trials were included in the study. The combination of cHT with RT and/or CT was found to improve both LC and OS in three out of five trials conducted in patients with cervical cancer. The analysis of both head and neck cancer trials demonstrated a marked increase in TR rates in cases where cHT was employed. Conversely, no such benefit was observed in non-small cell lung cancers and in a large multicenter cervical cancer trial. The treatment was well tolerated, with mild to moderate adverse events being the most common. CONCLUSION: It has been demonstrated that the combination of cHT with either RT or CT may improve TR and LC and, in certain cases, OS when compared with RT with or without CT alone. The occurrence of significant side effects is rare.
24 June 2026
Read appraisal →Acta orthopaedica
Registration and reporting characteristics of trials investigating exercise therapy following total knee arthroplasty: a systematic review.
BACKGROUND AND PURPOSE: Prospectively registering the primary trial outcome is important to reduce selective outcome reporting and increase the trustworthiness of findings, which guide clinical practice. The objectives of our systematic review were to explore and compare the reporting characteristics of prospectively and non-prospectively registered trials investigating exercise therapy following total knee arthroplasty. METHODS: Randomized trials comparing effects of exercise therapy after total knee arthroplasty for osteoarthritis were identified in 4 databases from 2000 to August 12, 2024. One primary outcome per trial was extracted, using a pre-specified hierarchical algorithm, irrespective of outcome domain. Pooled standardized mean differences (SMDs) were calculated on pre-specified outcome domains, and risk-of-bias assessed using the Cochrane Risk-of-Bias tool v2. RESULTS: 94 trials, comprising 9,396 participants, were included, of which 13 were prospectively registered, 33 retrospectively registered, and 48 unregistered. A single primary outcome was defined in 44% of the 94 trials, and 4 trials reported a primary outcome consistent with a prospective registration. The pooled SMD of primary outcomes was 0.06 (95% confidence interval [CI] -0.03 to 0.16) for prospectively registered trials, 0.67 (CI 0.22-1.11) for retrospectively registered trials, and 0.59 (CI 0.32-0.86) for unregistered trials. Lower risk-of-bias ratings and higher proportions of intention-to-treat adherence, dropout reporting, and adverse event reporting were observed among prospectively registered trials. CONCLUSION: Among prospectively registered trials we showed smaller effect size estimates between interventions with lower risk-of-bias ratings, and higher proportions of intention-to-treat adherence, dropout reporting, and adverse event reporting in contrast to trials without prospective registration; furthermore, clear specification of a single primary outcome was uncommon among trials evaluating exercise therapy after total knee arthroplasty.
23 June 2026
Read appraisal →Acta orthopaedica
Predicting persistent pain after total knee arthroplasty using different machine learning algorithms
BACKGROUND AND PURPOSE: After total knee arthroplasty (TKA), 10-20% of patients remain unsatisfied. Well-performing clinical prediction models can provide individualized risk estimates and stratification in terms of poor outcomes, resulting in unnecessary surgeries being avoided and patients being counseled preoperatively. We aimed to create a precise, well-performing prediction model for clinical application using different machine learning algorithms to predict those patients who will have residual pain, a low total Oxford Knee Score (OKS) and the patient group who do not achieve minimally clinical important difference (MCID) in OKS 1 year after TKA. METHODS: We conducted a retrospective cohort study based on patients who had undergone primary TKA at our institution combined with 751 patient-related variables. The multivariable models used were based on the results of univariate analysis. We used the machine learning method Extreme Gradient Boosting (XGBoost). The discrimination capability of the models was measured with the area under the curve (AUC). RESULTS: 11,755 patients were included in this study. There were 850 (7.2%) patients who experienced persistent pain 1 year after TKA. The AUC was 0.67. For the secondary outcomes, the AUC values were similar. The most important variables in the model were lower preoperative OKS, younger age, valgus malalignment, lower preoperative pain OKS, use of mild opioid, neuropathic pain medicine and thyroxine, and higher body mass index. CONCLUSION: The prediction models achieved poor AUCs. It seems clear that the prediction of pain and functional outcome after TKA is difficult, even with a large patient cohort combined with 751 patient-related variables and sophisticated machine-learning algorithms.
23 June 2026
Read appraisal →Medical science monitor : international medical journal of experimental and clinical research
Development and Validation of Machine-Learning-Based Prediction Models for Thyroid Diseases During Pregnancy
BACKGROUND International guidelines recommend early screening based on targeted risk factors to identify thyroid disease during pregnancy. The complexity of these risk factors makes accurate prediction challenging. This study aimed to develop and compare multiple machine-learning-based predictive models for thyroid disease during pregnancy. MATERIAL AND METHODS This retrospective study analyzed the clinical characteristics of 5461 women who gave birth at a single center. The dataset was divided into training and test sets. In the training set, feature variables associated with thyroid disease during pregnancy were selected using the Boruta algorithm. Eight models were developed: logistic regression, Bayesian approach, k-nearest neighbors, support vector machine, neural network, classification and regression tree, extreme gradient boosting, and random forest (RF). Model performance was evaluated using the receiver operating characteristic (ROC) curve, precision-recall curve (PRC), calibration curve, and decision curve analysis. RESULTS Nine feature variables were identified: age, height, pre-pregnancy weight, gravidity, parity, primiparity or multiparity, hypertensive disorders of pregnancy, scarred uterus, and autoimmune disease. The RF model demonstrated the best performance, with accuracy of 0.98387819 and 0.99597990, Matthews correlation coefficient of 0.96794139 and 0.97781292, log loss of 0.12670703 and 0.09442025, Brier score of 0.02495798 and 0.01921069, area under the ROC curve of 0.99877170 and 0.99991140, and area under the PRC of 0.99864486 and 0.99922572 in the training and test sets, respectively. CONCLUSIONS The RF model demonstrates excellent discriminative performance, accuracy, consistency, and generalizability in predicting thyroid disease during pregnancy.
23 June 2026
Read appraisal →Journal of neural engineering
Uncertainty-aware multi-path framework with dynamic arbitration for Alzheimer's disease MRI classification
Objective.As a progressive neurodegenerative disorder, Alzheimer's disease (AD) requires early and accurate diagnosis to delay pathological progression and improve clinical outcomes. Recently, Deep-learning methods based on structural MRI (sMRI) have been widely used for computer-aided AD diagnosis. However, reliable deployment remains challenging due to its sample-specific heterogeneity and the lack of explicit mechanisms for handling prediction uncertainty. In addition, common test-time perturbations and augmentation-based inference strategies could introduce anatomical inconsistency in medical images, leading to unstable decisions for hard cases.Approach.Accordingly, TriPathNet with Arbiter, as an uncertainty-aware multi-path framework, is proposed in the study. The framework learns spatially complementary representations via multi-view parallel encoding with metric-guided coordination. In addition, an augmentation-aware arbitration mechanism is introduced to adaptively correct predictions for high-uncertainty samples during inference.Main results.On the ADNI dataset for the AD vs. cognitively normal task, the proposed method achieves 98.85% accuracy, 99.18% sensitivity, 97.06% specificity, and 99.61% AUC, yielding satisfactory performance.Significance.The results indicate that TriPathNet with Arbiter provides an accurate and robust solution for sMRI-based computer-aided AD diagnosis, maintaining stable performance under test-time perturbations.
20 June 2026
Read appraisal →Journal of medical Internet research
The Open Syndrome Definition as a Machine-Readable Standard for Public Health: Design and Implementation Study
BACKGROUND: Case definitions are essential for effectively communicating public health threats. However, the absence of a standardized, machine-readable format poses significant challenges to interoperability, epidemiological research, data sharing, and the application of computational methods, including artificial intelligence. These barriers complicate collaboration across regions and organizations and hinder technological progress in public health. OBJECTIVE: This study aims to propose and release the first open, machine-readable format for representing case and syndrome definitions, together with tools and resources that enable their standardized and scalable use. METHODS: We developed the Open Syndrome Definition, a structured, machine-readable schema for representing case and syndrome definitions. We compiled official public health case definitions from multiple institutions and converted them into standardized, machine-readable representations using open-source tools. These tools, available through GitHub under the Massachusetts Institute of Technology license, automate the translation of narrative definitions into structured data. We also created a platform for browsing, analyzing, and contributing new definitions on our initiative website. RESULTS: The Open Syndrome Definition format enabled consistent, automated representation of case definitions across different diseases and jurisdictions. The conversion tools achieved high semantic fidelity, as assessed by qualitative expert review, between narrative and structured representations, supporting human verification and automated analysis. The dataset and accompanying tools demonstrated structural and semantic interoperability by standardizing definitions from various health systems into a unified format and integrating existing medical ontologies through JSON for Linked Data. To further illustrate practical applicability and downstream usage, we introduced a data filtering prototype that allows users to upload their own datasets and verify the results against the standardized definitions. CONCLUSIONS: The Open Syndrome Definition establishes a foundation for consistent and machine-readable public health definitions, facilitating reproducible research and interoperability at scale. By enabling systematic data exchange and artificial intelligence-driven analysis, it strengthens public health preparedness and supports more rapid, coordinated responses to emerging health threats.
19 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 →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
How Machine Learning Can Help Close Evidence Gaps for Drug Safety in Pregnant Women
Women, especially pregnant women, have historically been excluded from clinical research. In this News and Perspectives article, JMIR Correspondent Michelle Falci reports on how advances in data analytics may help bridge evidence gaps.
29 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 →BMJ paediatrics open
Enhancing respiratory virus surveillance among hospitalised children: a machine learning-based predictive model.
BACKGROUND: Viral respiratory tract infections (vRTIs) are a leading cause of paediatric hospitalisation and healthcare utilisation. Existing syndromic surveillance tools, including the WHO Severe Acute Respiratory Infection definition, demonstrate limited diagnostic accuracy in children whose symptom profiles vary widely. This study aimed to develop a machine learning (ML) model to predict microbiologically confirmed vRTIs in hospitalised children and to evaluate performance across age groups and viral pathogens. METHODS: We conducted a retrospective cross-sectional study of 2050 paediatric patients (<18 years) admitted with acute respiratory infections to two tertiary paediatric hospitals in Canada. Predictors included age, sex, hospital transfer status, chronic comorbidity status and 22 presenting symptoms. The primary outcome was microbiologically confirmed vRTI, determined by multiplex PCR or rapid antigen testing. Six ML algorithms were trained and the best-performing model, identified by area under the receiver operating characteristic curve (auROC), was tested on age subgroups, viral pathogens and sites. RESULTS: Among 2050 patients (median (IQR) age 2.4 (0.8-5.2) years), 1831 (89.3%) tested positive, most commonly for respiratory syncytial virus (RSV) (38.7%) and enterovirus/rhinovirus (32.8%). Logistic regression with L2 regularisation demonstrated the best performance (auROC, 0.754; 95% CI 0.697 to 0.808; sensitivity, 69.2%; specificity, 69.9%), with greatest performance among children <1 year (auROC, 0.763) and RSV cases (auROC, 0.727). CONCLUSIONS: An ML-based logistic regression model using admission data accurately predicted paediatric vRTIs, outperforming traditional syndromic surveillance definitions, especially among infants <1 year. By integrating ML models into hospital electronic medical records, healthcare systems can achieve enhanced respiratory virus surveillance, faster outbreak detection, greater diagnostic efficiency and improved pandemic preparedness.
23 May 2026
Read appraisal →BMJ (Clinical research ed.)
Concordance between target trial emulation and randomised controlled trials: systematic review and meta-analysis
OBJECTIVES: To assess the concordance of results between paired target trial emulations and their benchmarking randomised controlled trials and to explore determinants of emulation success. DESIGN: Systematic review and meta-analysis. DATA SOURCES: Medline, EMBASE, Scopus, PsycINFO, and Web of Science (2010 to 2025). STUDY SELECTION: Observational studies that explicitly aimed to emulate target randomised controlled trials of a health intervention were included. DATA EXTRACTION: Data were extracted on study domain, target population, intervention, comparator, and main outcomes for each emulation and corresponding randomised controlled trial. MAIN OUTCOME MEASURES: For 21 predefined emulation design features, concordance was assessed using Pearson correlation coefficients, standardised difference agreement, and ratio of ratios, with subgroup and regression analyses of study characteristics and concordance. RESULTS: Among 107 target trial emulation-randomised controlled trial pairs, the Pearson correlation coefficient was 0.59 (95% confidence interval (CI) 0.45 to 0.70), standardised difference agreement was 79% (85/107), and summary ratio of ratios was 0.96 (95% CI 0.92 to 1.01; I2=36%). In 63 pairs with closer emulation of trial design, a higher level of agreement was observed, with a Pearson correlation coefficient of 0.83 (0.73 to 0.89) and standardised difference agreement of 87% (55/63). In subgroup analysis, trial emulations tended to systematically underestimate treatment effects from randomised controlled trials for specific outcomes related to venous thromboembolism and major adverse cardiovascular events, while overestimating those for respiratory outcomes, with pooled ratio of ratios of 0.76 (0.58 to 1.00; I2=40%), 0.91 (0.86 to 0.96; I2=32%), and 1.20 (1.03 to 1.40; I2=40%), respectively. Moreover, emulations based on claims data tended to underestimate the treatment effects (ratio of ratios 0.90, 0.82 to 0.99; I2=38%). In univariate regression analyses, poorer target trial emulation-randomised controlled trial concordance was associated with emulation designs with imbalanced baseline population characteristics, treatment started in hospital, and lower outcome emulation quality. CONCLUSIONS: Current target trial emulations achieve moderate concordance when replicating their corresponding randomised controlled trials. Concordance could be improved by better emulation design, including better baseline and outcome emulation and leveraging multisource linked databases. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42024619811.
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 →Cancer causes & control : CCC
Prediction models for early detection and diagnosis of lung cancer in people who have never smoked: a systematic review and critical appraisal
BACKGROUND: Never-smokers can develop lung cancer, but this possibility is often overlooked by patients and physicians. We conducted a systematic review to identify existing prediction models that could be implemented in primary care or at a population level to facilitate early detection and diagnosis of lung cancer in never-smokers. METHODS: This study was registered on PROSPERO (ref: CRD42023374471). We systematically searched literature on the Medline, Embase, PsycINFO, and CINAHL databases published before 22 January 2025, with additional hand searching. Primary care or population-level data from subjects including never-smokers had to be used for model derivation. Studies involving specialized tests (radiological or genetic) were excluded. We used CHARMS to guide data extraction and critical appraisal, the TRIPOD statement for model evaluation, and PROBAST for risk of bias assessment. RESULTS: Among 2,431 studies retrieved, 31 models were included. Eight models were developed exclusively for never-smokers, but none were at low risk of bias. Among 23 models derived from never- and ever-smokers, five were at low risk of bias. Two were diagnostic models with 1-2 years prediction horizons, and three were prognostic models with 2-10 years prediction horizons. Methodological issues from the included studies were identified, analyzed, and discussed. CONCLUSION: This systematic review critically appraises and summarizes key information from currently available prediction models for lung cancer in never-smokers. The findings can inform future research to improve care and services for this underserved population.
20 May 2026
Read appraisal →Lupus science & medicine
Incidence of serious infections, herpes zoster and hepatitis B virus reactivation among patients with SLE: a cohort study using a health claims database in Japan
OBJECTIVES: To investigate the real-world incidence of serious infections (SI), herpes zoster (HZ) and hepatitis B virus (HBV) reactivation among patients with SLE. METHODS: This retrospective, non-interventional cohort study used an insurance claims database in Japan to calculate the incidence of SI requiring hospitalisation, HZ and HBV reactivation in patients with SLE from 1 April 2017 to 31 March 2021. As an exploratory analysis, HRs and 95% CIs were estimated for clinically relevant patient characteristics for SI and HZ, adjusting for age and sex. RESULTS: Among 10 865 patients with SLE (female: 83.1%, median age: 53.0 years), including those who had previously experienced each complication, the incidence rates (per 100 patient-years) for SI, HZ and HBV reactivation were 4.80 (95% CI 4.53 to 5.08), 0.52 (0.44 to 0.61) and 0.37 (0.30 to 0.45), respectively. For those who had no previous history of each complication, incidence rates were 3.67 (3.44 to 3.92), 0.48 (0.40 to 0.58) and 0.25 (0.20 to 0.32), respectively. Incidence rates in specific subgroups were as follows: history of each complication (SI: 37.93, HZ: 4.19, HBV reactivation: 72.01), history of SI (HZ: 1.51, HBV reactivation: 1.75) and glucocorticoid pulse therapy during follow-up (SI: 64.23, HZ: 3.10, HBV reactivation: 4.85). CONCLUSIONS: These results provide data on the occurrence of selected infections in SLE and may inform safety management.
18 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 →BMC medical research methodology
When to stop reviewing: validation of stop criteria in ASReview.
BACKGROUND: Systematic reviews are essential for evidence-based research, but they often require a great deal of time and effort. Although title and abstract (T&A) screening is just one part of the review process, it can be very time-consuming when search strategies retrieve large numbers of records. Given the exponential growth of scientific publications in recent decades, tools such as ASReview, which use machine learning (ML) for active learning-based screening, aim to reduce the workload. However, since ASReview helps users prioritise potentially relevant studies rather than supporting them in screening all records, a key question arises: at what point can the screening process safely stop without risking the omission of relevant studies when the entire dataset is not being reviewed? METHODS: This simulation study tested three proposed stop criteria for terminating screening in ASReview without loss of relevant data: (1) stopping after a calculated number of relevant studies based on an initial sample; (2) stopping after a fixed number of consecutively studies deemed irrelevant; (3) stopping after a predefined percentage of the dataset has been screened. A total of 35,000 automated title and abstract screenings were conducted using five datasets from the SYNERGY repository. Key outcomes included the percentage of studies screened until the last relevant study was found and the number of relevant studies missed under each stop criterion. RESULTS: The proportion of the dataset that needed to be screened to identify all relevant studies (as pre-classified in the SYNERGY dataset) varied greatly across datasets, ranging from 2.9% to 76.9% on average. None of the tested stop criteria could consistently identify all relevant studies across all datasets. Stop criterion 1 was reliable in only 2% of simulations. Stop criterion 2 showed high variability, with thresholds ranging from 2% to 61%, depending on the dataset. Stop criterion 3 failed to define a universal percentage applicable across datasets. CONCLUSIONS: ASReview can reduce screening workload by prioritizing potentially relevant studies through ML-based ranking, thereby allowing researchers to identify relevant studies earlier in the screening process. However, no stop criterion reliably ensures that all relevant studies are identified. Early stopping may result in missed studies, depending on dataset characteristics. Current stop criteria should be applied cautiously and potentially combined with quality assurance measures. Further research is needed to develop more robust and generalizable stopping rules. TRIAL REGISTRATION: Not applicable - this is a simulation study, not a registered systematic review.
13 May 2026
Read appraisal →The aging male : the official journal of the International Society for the Study of the Aging Male
SPISE index and ensemble machine learning refine cardiovascular risk stratification in stage 0-3 CKM syndrome
BACKGROUND: While the single-point insulin sensitivity estimator (SPISE) shows promise as an insulin resistance biomarker, its association with cardiovascular disease (CVD) in early CKM stages (0-3) remains underexplored. METHODS: We analyzed 6480 participants with CKM stage 0-3 from the China Health and Retirement Longitudinal Study. CVD outcomes were assessed relative to SPISE index levels. An ensemble machine learning model was employed to predict CVD risk. RESULTS: 6480 subjects were enrolled, of whom 967 developed CVD. After stratifying participants into SPISE quartiles (Q1-Q4) and adjusting for covariates, higher quartiles were linked to a lower CVD risk. This study developed an LR+GMM (Logistic Regression + Gaussian Mixture Model) ensemble model to predict CVD risk using five strong predictors: SPISE, high-density lipoprotein cholesterol (HDL-c), diastolic blood pressure (DBP), body mass index (BMI), and glycated hemoglobin (HbA1c). The model performed well, achieving an accuracy (ACC) of 0.986 and an area under the receiver operating characteristic curve (AUC) of 0.932. CONCLUSIONS: The SPISE index is a significant inverse predictor of CVD risk in individuals with stage 0-3 CKM syndrome. The LR+GMM ensemble model, incorporating the SPISE index and four clinical metrics, demonstrated outstanding predictive performance.
12 May 2026
Read appraisal →The Journal of asthma : official journal of the Association for the Care of Asthma
Prediction model for cardiovascular mortality risk in asthma patients based on machine learning
BACKGROUND: Asthma is a common chronic respiratory disease, cardiovascular disease (CVD) mortality constitutes a major public health concern. At present, no personalized tools are available to accurately estimate CVD mortality risk in this population. Accordingly, a machine learning model was developed to predict incident CVD mortality of asthma. METHODS: Data were extracted from NHANES (2007-2012). Predictors were selected by LASSO and multivariable Cox regression. Discrimination and calibration were quantified with ROC curves and calibration curves, nomogram and web calculator were subsequently deployed. Incremental predictive value relative to ASCVD risk and Framingham risk scores were assessed with the integrated discrimination improvement and net reclassification improvement. Six machine learning algorithms were trained, and feature importance of the optimal model was interpreted with SHAP values. RESULTS: Among 2,033 adults participants with asthma, with mean age 44 years, and they comprised 834 males (41.02%) and 1199 females (58.98%). Twelve variables (age, sex, smoking, hypertension, diabetes, CKD, SBP, LYM, UA, K, FENO, and PEF) were retained as predictors of CVD mortality. Calibration was stable across training sets (C-index = 0.863) and validation sets (C-index = 0.832). The pulmonary function test augmented NHANES model outperformed both ASCVD risk and Framingham risk scores. RF and XGB model achieved the highest discrimination. CONCLUSION: A machine learning model integrating routine clinical and pulmonary function test indices was developed. The model enables accurate estimation of long term CVD mortality risk with asthma, facilitating early identification of high risk individuals and informing personalized preventive strategies.
9 May 2026
Read appraisal →Research synthesis methods
Transforming evidence synthesis: A systematic review of the evolution of automated meta-analysis in the age of AI
Exponential growth in scientific literature has heightened the demand for efficient evidence-based synthesis, driving the rise of the field of automated meta-analysis (AMA) powered by natural language processing and machine learning. This PRISMA systematic review introduces a structured framework for assessing the current state of AMA, based on screening 13,216 papers (2006-2024) and analyzing 61 studies across diverse domains. Findings reveal a predominant focus on automating data processing (52.5%), such as extraction and statistical modeling, while only 16.4% address advanced synthesis stages. Just one study (approximately 2%) explored preliminary full-process automation, highlighting a critical gap that limits AMA's capacity for comprehensive synthesis. Despite recent breakthroughs in large language models and advanced AI, their integration into statistical modeling and higher-order synthesis, such as heterogeneity assessment and bias evaluation, remains underdeveloped. This has constrained AMA's potential for fully autonomous meta-analysis (MA). From our dataset spanning medical (67.2%) and non-medical (32.8%) applications, we found that AMA has exhibited distinct implementation patterns and varying degrees of effectiveness in actually improving efficiency, scalability, and reproducibility. While automation has enhanced specific meta-analytic tasks, achieving seamless, end-to-end automation remains an open challenge. As AI systems advance in reasoning and contextual understanding, addressing these gaps is now imperative. Future efforts must focus on bridging automation across all MA stages, refining interpretability, and ensuring methodological robustness to fully realize AMA's potential for scalable, domain-agnostic synthesis.
6 May 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 →Renal failure
Decoding the renal-cochlear axis: explainable machine learning and phenotype clustering reveal high-risk hearing loss subtypes in CKD
This study develops a dual-level machine learning framework for risk stratification and phenotyping of hearing loss (HL) in patients with chronic kidney disease (CKD) using data from the National Health and Nutrition Examination Survey (NHANES). From a cohort of 3,402 CKD patients, feature selection via univariate and multivariate logistic regression identified key predictors, which were used to construct predictive models with nine machine learning algorithms. The eXtreme Gradient Boosting (XGBoost) model demonstrated superior performance, with mean area under the curve (AUC) values of 0.984 (training), 0.984 (validation), and 0.939 (testing). SHapley Additive exPlanations (SHAP) interpretation identified age as the predominant risk determinant. Subsequent Gaussian mixture modeling (GMM) clustered patients into two distinct subtypes: a low-risk subgroup (n = 1,075) with a 1.58% HL prevalence and a high-risk subgroup (n = 2,316) characterized by older age, elevated blood urea nitrogen and bicarbonate levels, and a 48.2% HL prevalence. A classifier trained on these subtypes achieved discrimination (AUC = 0.99974). A clinically a web-based tool was also developed based on the six most influential features. The findings establish a dual-level predictive framework integrating explainable ML and unsupervised clustering for HL risk assessment in CKD. This approach provides a robust strategy for the precision screening of high-risk subpopulations and supports the integration of hearing assessments into routine CKD.
24 Apr 2026
Read appraisal →Journal of the American College of Cardiology
Risk-Guided Atrial Fibrillation Screening With Artificial Intelligence-Enabled Electrocardiogram Models: A VITAL-AF Trial Analysis
BACKGROUND: Screening for atrial fibrillation (AF) may lead to earlier detection and initiation of preventive measures. Current AF screening approaches using a guideline age-based threshold of ≥65 years have shown limited yield. OBJECTIVES: In an AF screening trial, we assessed whether the screening effect was larger among individuals at elevated AF risk using validated clinical and electrocardiogram (ECG)-based artificial intelligence (AI) risk models. METHODS: VITAL-AF was a cluster-randomized trial of patients aged ≥65 years treated at 1 of 16 primary care practices affiliated with Massachusetts General Hospital. Patients randomized to a screening practice were screened using a single-lead ECG. Among VITAL-AF participants without prevalent AF with at least one 12-lead ECG within 3 years before enrollment, we estimated AF risk using 3 validated models derived outside of VITAL-AF: the Cohorts of Heart and Aging Research in Genomic Epidemiology-AF (CHARGE-AF) clinical score, an AI-based model using a 12-lead ECG alone (ECG-AI), and a model combining ECG-AI and CHARGE-AF (CH-AI). Two-year incident AF discrimination was assessed by the time-dependent area under the receiver-operating characteristic curve (AUROC) and average precision. AF screening effect was defined as the difference in 2-year incident AF diagnosis rate (per 100 person-years) in screening vs control across AF risk deciles. RESULTS: Of 30,630 VITAL-AF participants without prevalent AF, 16,937 had pretrial ECG and clinical data. Each score discriminated 2-year AF risk according to AUROC (CHARGE-AF: 0.711 [95% CI: 0.671-0.749]; ECG-AI: 0.784 [95% CI: 0.743-0.819]; CH-AI: 0.788 [95% CI: 0.754-0.824]) and average precision (0.0952 [95% CI: 0.0836-0.112]; 0.132 [95% CI: 0.113-0.157]; 0.133 [95% CI: 0.117-0.159]). An AF screening effect was observed in the top decile of CH-AI (AF diagnosis rate in screening 10.07/100 person-years [95% 8.28-11.87] vs 7.76 [95% 6.30-9.21] in control, P < 0.05), corresponding to a difference in AF diagnosis rate of 2.32/100 person-years (95% CI: 0.01-4.63) and number-needed-to-screen of 43 per year. CONCLUSIONS: Use of ECG-based AI and clinical factors identified individuals at particularly high risk for AF who may benefit from screening. Findings suggest a trade-off between increasing AF screening efficiency and decreasing population coverage (ie, restriction of the screening pool). Future studies are needed to determine whether a risk-based approach is optimal or whether consideration of additional clinical- and systems-level factors (eg, access, health care system engagement) can further refine AF screening strategies. (Screening for Atrial Fibrillation Among Older Patients in Primary Care Clinics [VITAL-AF]; NCT03515057).
16 Apr 2026
Read appraisal →European journal of orthopaedic surgery & traumatology : orthopedie traumatologie
Predicting minimal clinically important difference after hip arthroscopy: logistic regression versus machine learning
PURPOSE: Hip arthroscopy outcomes for femoroacetabular impingement (FAI) remain heterogeneous, with 30-40% of patients failing to achieve meaningful improvement. We aimed to develop and validate a predictive model for achievement of minimal clinically important difference (MCID) using routine preoperative variables, and to compare logistic regression with machine learning algorithms. METHODS: We retrospectively analyzed 310 consecutive patients undergoing hip arthroscopy for FAI (2010-2024) at a single institution. Predictor variables included age, sex, body mass index (BMI), morphological etiology, preoperative pain (VAS), and Hip Outcome Score-Activities of Daily Living (HOS-ADL). The outcome was achievement of MCID (≥ 9-point improvement in HOS-ADL). We developed logistic regression, random forest, and gradient boosting models. Internal validation used 10-fold stratified cross-validation. Temporal external validation used 80% of patients for training and 20% for testing. RESULTS: Mean age was 39.7 ± 12.0 years; 59.7% were male. MCID was achieved by 213 patients (68.7%). Lower preoperative HOS-ADL (OR 0.911, 95% CI 0.883-0.940, p < 0.001) and lower BMI (OR 0.846, 95% CI 0.771-0.928, p < 0.001) independently predicted MCID achievement. Logistic regression achieved AUC 0.804 (95% CI 0.76-0.85) on internal validation and AUC 0.819 on temporal external validation (accuracy 81.4%). Logistic regression outperformed random forest (AUC 0.787) and gradient boosting (AUC 0.777). Calibration was excellent. CONCLUSION: A parsimonious logistic regression model using routine preoperative variables demonstrated excellent and externally validated discriminative ability for predicting MCID after hip arthroscopy. The model outperformed machine learning algorithms while maintaining superior interpretability. The developed nomogram provides a practical tool for individualized outcome prediction.
9 Apr 2026
Read appraisal →European journal of pediatrics
Predicting coronary artery abnormalities in Kawasaki disease: Model development and validation
Routine echocardiography at one month after diagnosis is the current standard for screening coronary artery abnormalities (CAA), a major complication of Kawasaki disease. The study aimed to develop and validate models to predict CAA and assess whether routine echocardiography could be safely reduced in low-risk patients. Two prospective multicenter Japanese registries were utilized: PEACOCK (development/internal validation) and Post-RAISE (external validation). Variables obtained within one week of diagnosis were used to predict CAA at one month after diagnosis, defined as a maximum coronary artery Z score (Zmax) ≥ 2. The models included simple models using the previous maximum Z score only, logistic regression models, and machine learning models (LightGBM and XGBoost). Discrimination, calibration, and clinical utility were assessed. Among 4,973 PEACOCK and 2,438 Post-RAISE patients, the CAA incidence was 5.5% and 6.8%, respectively. Twenty-two models were developed using 29 variables. For external validation, a simple model using the maximum Z score at week 1 produced an area under the curve (AUC) of 0.79; adding other variables or using more complex models did not increase the AUC by more than 0.02. The models failed to efficiently reduce the number of echocardiographic examinations while minimizing missed cased of CAA.Conclusion: Until superior predictors are identified, routine echocardiography at one month after diagnosis should remain the standard practice.
9 Apr 2026
Read appraisal →Journal of the American Heart Association
Development and Validation of a Machine Learning Model for Incident Heart Failure Prediction in Chronic Kidney Disease: A Multicenter Cohort Study
BACKGROUND: Chronic kidney disease (CKD) and heart failure (HF) share pathophysiological mechanisms, rendering HF one of the most burdensome cardiovascular complication in CKD. Current HF prediction models, derived from the general population, exhibit limited accuracy in CKD, thus necessitating a CKD-specific risk model and clinical implementation. METHODS: The development set comprised 52 251 patients with CKD from the China Renal Data System (70% training; 30% internal validation). External validation used 21 798 patients from independent Chinese hospitals and 3323 UK Biobank participants. Outcome was 5-year new-onset HF. Five machine learning models were developed, with performance assessed via area under the curve and compared using DeLong test. The top-performing extreme gradient boosting model was simplified via forward stepwise selection; feature importance quantified using Shapley additive explanations. RESULTS: In the Chinese external validation cohort, the extreme gradient boosting model outperformed others (area under the curve, 0.879 [95% CI, 0.871-0.887]; DeLong P<0.05), with an area under the curve of 0.851 (95% CI, 0.831-0.870) in the UK Biobank cohort. The model was simplified to 9 predictors; Shapley additive explanations analysis ranked estimated glomerular filtration rate and albuminuria as among the most important features. In Chinese and UK Biobank external validation sets, the simplified extreme gradient boosting model showed a ΔC-statistic of 0.050 (0.045-0.056) and 0.006 (-0.013-0.024) versus the Atherosclerosis Risk in Communities risk score, and 0.194 (0.183-0.204) and 0.145 (0.107-0.182) versus the Predicting Risk of Cardiovascular Disease Events Equation, respectively. A web-based calculator was deployed (https://clinician.shinyapps.io/HF_Risk_Predictor_for_CKD/). CONCLUSIONS: The 9-variable extreme gradient boosting model tailored for patients with CKD may help predict HF risk in this high-risk population. Validation across diverse populations is warranted to confirm its generalizability.
8 Apr 2026
Read appraisal →Medicine
Development of a prediction model for the risk of prolonged ICU length of stay in patients undergoing coronary artery bypass surgery: A retrospective analysis based on the MIMIC-IV database
The duration of hospitalization, particularly in the intensive care unit (ICU), for patients undergoing coronary artery bypass grafting (CABG), is influenced by both patient prognosis and treatment-related costs. Reducing the ICU length of stay (LOS) in CABG patients is critical for improving healthcare resource utilization. This study aimed to develop a model for identifying risk factors associated with prolonged ICU LOS in CABG patients, thus providing a foundation for clinical treatment, healthcare safety, and quality management research. In this single-center, retrospective cohort study, data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database were analyzed. Clinical data from 6152 CABG patients who required treatment in ICU were included. The dataset was split into a training set and an internal validation set. Predictor variables were identified using least absolute shrinkage and selection operator regression. A nomogram prediction model was developed based on the identified predictors. The predictors incorporated into the model included the Charlson comorbidity index, acute physiology score III, atrial fibrillation, heart failure (HF), body mass index, acute kidney injury stage, type 1 diabetes mellitus and glucose levels. The prediction model demonstrated robust performance, achieving an area under the curve of 0.789 in the training cohort and 0.790 in the validation cohort. The Hosmer-Lemeshow (H-L) test and calibration plots validated the model's accuracy in both cohorts. The nomogram prediction model proposed in this study provides significant clinical utility for predicting prolonged ICU LOS in patients undergoing CABG. This model can assist clinicians in identifying CABG patients at risk for prolonged ICU LOS, thus enabling timely interventions and improving prognostic outcomes.
5 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 →Journal of advanced research
Machine learning ensemble meets clinical practice: developing a real-world risk prediction model for metabolic syndrome using super learner and scorecard approaches
INTRODUCTION: Metabolic syndrome (MetS) refers to a cluster of metabolic abnormalities that significantly increase the risk of developing cardiovascular diseases (CVDs). Traditional binary definitions of MetS fall short in capturing its severity spectrum, limiting personalized risk stratification. OBJECTIVES: We aimed to develop a super learner model and multilevel risk scorecard to improve MetS risk prediction and support early cardiovascular risk identification. METHODS: This study included a total of 460,256 health examination records from Zhejiang, China (2018-2023), with 344,925 used for model development and 115,331 used for external validation. A super learner model combining multiple machine learning algorithms was constructed to predict MetS risk. Key predictors identified through feature selection were used to develop a logistic regression-based MetS risk scorecard, stratifying individuals into five risk levels for practical and interpretable applications. RESULTS: The super learner model achieved area under the receiver operating characteristics (AUCs) of 0.816 (95 % confidence interval [CI]: 0.814-0.817) and 0.810 (95 % CI: 0.808-0.813) in the development and external validation cohorts, respectively. The risk scorecard, which incorporates ten predictors, demonstrated comparable performance, with AUCs of 0.793 (95 % CI: 0.791-0.794) and 0.788 (95 % CI: 0.785-0.791) in the respective cohorts. Based on the risk scorecard, individuals were stratified into five categories: very low, low, normal, high, and very high risk levels. CONCLUSION: The super learner model provides a highly accurate tool for MetS risk prediction, whereas the risk scorecard offers a practical and interpretable solution for clinical and personal use. These models enable precise risk assessment to guide prevention and improve outcomes.
4 Apr 2026
Read appraisal →