Frontiers in digital health最新文献

{"title":"Correction: Feasibility study of noninvasive assessment of emotional state in patients with ALS.","authors":"Marc Garbey, Quentin Lesport, Gülşen Öztosun, Megan Heidebrecht, Kamyab Pirouz, Elham Bayat","doi":"10.3389/fdgth.2026.1977072","DOIUrl":"https://doi.org/10.3389/fdgth.2026.1977072","url":null,"abstract":"<p><p>[This corrects the article DOI: 10.3389/fdgth.2026.1886561.].</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1977072"},"PeriodicalIF":5.2,"publicationDate":"2026-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13643689/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149139204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of a machine learning model to identify individuals with VCTE-derived at-risk MASH in a Spanish Mediterranean region.","authors":"Marta Cedenilla, David Marti-Aguado, Josep Redon, Inma Sauri-Ferrer, Jorge Navarro, Desamparados Escudero, Anton Gomez, Luis Cea-Calvo, Carlos J Martos-Rodriguez, Javier Díaz Carnicero","doi":"10.3389/fdgth.2026.1829881","DOIUrl":"10.3389/fdgth.2026.1829881","url":null,"abstract":"<p><strong>Introduction: </strong>While current guidelines recommend screening for liver disease in target populations, existing non-invasive tests have limitations in identifying at-risk metabolic dysfunction-associated steatohepatitis (MASH).</p><p><strong>Materials and methods: </strong>This retrospective study used two large datasets from the general population in the Valencian Community region (Spain). The primary goal was to develop a machine learning model to identify individuals with at-risk MASH. The model was constructed using a well-characterized dataset (<i>n</i> = 2,267) with available vibration-controlled transient elastography (VCTE). Supervised machine learning models (Random Forest and XGBoost) were trained using a 10-fold cross-validation to classify at-risk MASH individuals, defined as those with a controlled attenuation parameter (CAP) ≥ 275 dB/m together with a liver stiffness measurement (LSM) > 8 kPa and/or FibroScan-AST score (FAST) > 0.35. The final model, named VARM-7 (Valencia At-Risk MASH), comprised age and six commonly available laboratory variables. VARM-7 was sequentially applied to the Valencian public healthcare database (<i>n</i> = 3,411,069) to estimate the proportion of at-risk MASH individuals in the overall population, and in target populations with type 2 diabetes mellitus (T2DM) and obesity.</p><p><strong>Results: </strong>In the independent validation dataset, VARM-7 exhibited a significantly higher AUROC compared with Fibrosis-4 (FIB-4) for identifying at-risk MASH (0.84 <i>vs</i>. 0.64, respectively, <i>p</i> < 0.001). When sequentially applied to the overall Valencian Community population, 11.7% of individuals were estimated to have at-risk MASH, with increasing estimates up to 29.4% and 44.0% in subjects with obesity and T2DM, respectively.</p><p><strong>Conclusion: </strong>VARM-7 showed strong performance in identifying individuals with a VCTE-derived at-risk MASH. Our model could improve disease screening and referral pathways, but external validation and prognostic evaluation are needed before its implementation.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1829881"},"PeriodicalIF":5.2,"publicationDate":"2026-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13638652/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Discrepancy between medical knowledge and clinical reasoning: evaluating large language models in prenatal diagnosis.","authors":"Qiuling Guo, Yanyun Du, Jingcheng Dai, Weiru Lin","doi":"10.3389/fdgth.2026.1917573","DOIUrl":"10.3389/fdgth.2026.1917573","url":null,"abstract":"<p><strong>Purpose: </strong>Large language models (LLMs) have demonstrated impressive performance across various medical examinations; however, their competency in subspecialty clinical reasoning, particularly in prenatal diagnosis, remains poorly characterized.</p><p><strong>Materials and methods: </strong>A cross-sectional comparative study was conducted by using 125 questions from China's nationally standardized Prenatal Ultrasound Screening Qualification Examination. Three LLMs and two human physicians were evaluated. Questions were classified along three independent taxonomic axes: modality, disease system, and question format. Cochran's <i>Q</i> test was used for omnibus comparisons, with <i>post-hoc</i> pairwise McNemar tests where were significant.</p><p><strong>Results: </strong>Overall accuracy did not differ significantly among the five test-takers (<i>p</i> = 0.126): the chief physician scored highest (89.60%, 112/125), followed by Claude-Sonnet-4.6 (85.60%, 107/125), Grok-4 (84.80%, 106/125), ChatGPT-5 (84.00%, 105/125), and the attending physician (80.00%, 100/125). No significant differences were observed when stratified by modality (text-only, <i>p</i> = 0.146; imaging diagnostic questions, <i>p</i> = 0.603) or by any individual disease system (all <i>p</i> > 0.05). However, performance divergence emerged sharply across question types. In clinical vignette-based questions (<i>n</i> = 20), the chief physician (95.00%) significantly outperformed all other test-takers (<i>p</i> = 0.016), with a 25% gap over the best-performing LLM (Grok-4, 70.00%). In sequential item sets, Grok-4 (95.74%) and the chief physician (93.62%) both significantly outperformed the attending physician (76.60%; both <i>p</i> < 0.05). Conversely, LLMs excelled in knowledge-retrieval tasks, with ChatGPT-5 and Claude-Sonnet-4.6 both achieving 100% accuracy in matching questions. No significant differences were detected among test-takers for factual recall, matching, or multiple-choice multiple-answer questions (all <i>p</i> > 0.05).</p><p><strong>Conclusion: </strong>LLMs are well-positioned to serve as powerful knowledge augmentation tools, but are not yet ready to function as autonomous clinical reasoners in this complex, ethically consequential subspecialty.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1917573"},"PeriodicalIF":5.2,"publicationDate":"2026-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13640009/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119454","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SOMA: a preventive health governance framework for adaptive, consumer-facing digital health guidance.","authors":"Michael Robert Mueller","doi":"10.3389/fdgth.2026.1902578","DOIUrl":"10.3389/fdgth.2026.1902578","url":null,"abstract":"<p><p>This paper introduces SOMA, a conceptual preventive health governance framework for adaptive, consumer-facing digital health guidance, designed to support everyday preventive health self-governance. Non-communicable and lifestyle-sensitive chronic diseases account for the majority of global mortality, yet existing consumer-facing digital health tools rarely provide coherent, governed preventive guidance that integrates multimodal personal health data, scientific evidence, user-defined goals, real-world feasibility, relational support, and explicit safety constraints within a unified architecture. SOMA conceptualizes preventive health self-governance as the iterative, user-governed process through which individuals interpret their evolving health state, define goals and feasibility constraints, receive evidence-informed guidance, and act within boundaries that reduce the risk of unsafe or inappropriate digital health guidance. The framework theorizes an architecture of five interdependent components: a Digital Twin Core synthesizing a multimodal Health State Vector across biological, wearable, subjective, behavioral, and contextual layers; a Scientific Intelligence Core evaluating evidence quality, uncertainty, and applicability; a Relational Avatar Core supporting disclosure, trust calibration, and user-facing communication; a Personal Health Governance Profile embedding user-defined goals, feasibility constraints, and autonomy preferences; and a Continuous Learning Engine updating the model based on outcomes, evidence, and user feedback, all bounded by non-negotiable Safety and Medical Escalation Boundaries. The theoretical contribution lies in integrating previously siloed approaches (digital health twin logic, evidence-weighted scientific intelligence, relational guidance, behavioral self-governance theory, and personal health governance) into a unified preventive architecture whose governance properties emerge from the structured interdependence of its components. Five testable hypotheses are derived from the architecture to guide a staged empirical validation programme. SOMA is a conceptual framework that has not been empirically validated and makes no claims regarding clinical effectiveness, disease prevention, or deployment readiness.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1902578"},"PeriodicalIF":5.2,"publicationDate":"2026-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13638536/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119504","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The precision ceiling: information-theoretic accuracy bounds for responsible AI health screening.","authors":"Kai Ding, Zhi Li, Xiang Zou, Jingdong Jia, Chenlin Li, Zijing Jiang","doi":"10.3389/fdgth.2026.1956262","DOIUrl":"https://doi.org/10.3389/fdgth.2026.1956262","url":null,"abstract":"<p><p>Large language models pass licensing examinations and foundation models match human diagnostic accuracy, yet the sensing modalities that supply these algorithms with data obey unchanged physical limits. This article introduces the <i>precision ceiling</i>: the maximum accuracy attainable by any model operating on a given sensing modality, bounded by the information-theoretic capacity of the measurement channel rather than by algorithmic capacity. Grounded in the data processing inequality, the ceiling is an <i>a priori</i> constraint whose clinical magnitude must be empirically characterized per modality. We develop a three-layer nested constraint taxonomy (Physical Law <math><mo>→</mo></math> Physiological Variability <math><mo>→</mo></math> Modality Gap), demonstrate the phenomenon across four optical and photoelectric sensing applications (postural screening, wearable photoplethysmography, fundus imaging, and dermatological AI), and propose two conceptual design contributions-the IICRC design principles (Identify, Internalize, Communicate, Recognize, Cascade) and the Precision Ceiling Audit, a qualitative pre-deployment characterization template-offered for future empirical validation. The framework does not replace existing responsible-AI initiatives; it identifies where humility and safety are physically inevitable rather than aspirational, providing a principled basis for modality-specific accuracy declarations, intended-use restrictions, and escalation protocols aligned with emerging regulatory expectations.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1956262"},"PeriodicalIF":5.2,"publicationDate":"2026-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13635159/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119431","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From 5G smart hospitals to 6G-enabled learning hospitals: a governance-oriented framework for healthcare technology, policy, and responsible deployment.","authors":"Guiyang Zhou, Lisha Wu","doi":"10.3389/fdgth.2026.1923445","DOIUrl":"https://doi.org/10.3389/fdgth.2026.1923445","url":null,"abstract":"<p><strong>Objectives: </strong>Discussion of 6G in healthcare often moves directly from anticipated network capabilities to proposed clinical applications, although application-level evidence, current connectivity alternatives, and 6G-specific evidence differ substantially in maturity. This narrative review develops an evidence-calibrated, governance-oriented framework for prospective 6G-enabled learning hospitals and examines when future 6G may add meaningful value beyond existing infrastructure.</p><p><strong>Methods: </strong>A multidisciplinary narrative synthesis covered smart hospitals, learning health systems, 5G/6G communication, IoMT, edge-cloud computing, AI/XAI, robotics, digital twins, cybersecurity, interoperability, implementation, and digital-health governance. Structured searches of PubMed, Web of Science, IEEE Xplore, and Google Scholar covered 1 January 2010-31 January 2026, supplemented by relevant standards, regulatory and policy sources, and targeted recent studies. Evidence was interpreted using an author-developed Evidence Maturity Classification (EMC) spanning demonstrated implementation, clinically grounded empirical, technical/benchmark, and prospective evidence.</p><p><strong>Results: </strong>The synthesis produced a seven-layer architecture, nine use-case analyses, a capability and evidence-maturity matrix, a twelve-domain risk-governance-metric framework with decision gates, a five-phase implementation roadmap, and an executive checklist. Across use cases, evidence for the underlying healthcare application was more mature than evidence for incremental benefit specifically attributable to 6G. Existing wired networks, Wi-Fi, private cellular systems, 5G/5G-Advanced, and edge-cloud infrastructure already support many monitoring, telemedicine, imaging, home-care, logistics, maintenance, and selected teleoperation functions. Future 6G is therefore most relevant when meaningful residual capability gaps remain, particularly combinations of mobility continuity, predictable end-to-end service, dense device coordination, multimodal synchronization, integrated sensing or positioning, and communication-computing orchestration.</p><p><strong>Conclusions: </strong>6G adoption should follow demonstration of clinical or operational value, a residual capability gap, and governance readiness. Hospitals should use the least complex infrastructure that meets requirements, generate evidence through risk-proportionate pilots, and scale advanced connectivity only when incremental value is demonstrated against optimized current alternatives. The framework provides a structured basis for these decisions but requires prospective validation across diverse hospital settings.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1923445"},"PeriodicalIF":5.2,"publicationDate":"2026-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13635659/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119414","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Impact of a digital mindfulness and motor imagery intervention on pain and perceived function in 203 patients with musculoskeletal pain.","authors":"Guillaume Zunzarren, Bertrand Garet, Sonja Cabarkapa, Jérôme Murgier","doi":"10.3389/fdgth.2026.1895131","DOIUrl":"https://doi.org/10.3389/fdgth.2026.1895131","url":null,"abstract":"<p><strong>Background: </strong>Musculoskeletal pain (MSP) is associated with persistent pain, functional impairment, and altered neuromuscular control mechanisms. Mindfulness and motor imagery interventions may target central sensorimotor dysfunctions involved in pain persistence. This study evaluated the effects of a digital intervention combining mindfulness and guided motor imagery on pain and perceived function in patients with MSP.</p><p><strong>Methods: </strong>This observational study analyzed 203 consecutive users with complete datasets who used a standardized digital intervention delivered through the Yuzit® application. The program combined mindfulness exercises and guided motor imagery sessions adapted to the painful anatomical area, covering eight musculoskeletal regions. Outcomes were assessed at baseline (T0), before the third session (T1), and after program completion (T2). Pain intensity was measured using the 11 points Numerical Rating Scale (NRS; 0-10) and perceived function using the Single Assessment Numeric Evaluation (SANE). The Minimal Clinically Important Difference (MCID) was defined as a VAS reduction ≥ 2 points. Secondary analyses evaluated the influence of age, sex, symptom duration, pain location, and baseline pain severity.</p><p><strong>Results: </strong>The cohort included 122 women (60.1%) and 81 men (39.9%) with a mean age of 46.4 ± 16.6 years. Mean NRS improved from 5.17 ± 2.00 at T0 to 3.26 ± 2.40 at T2 (<i>Δ</i> = -1.91; t = -12.45; <i>p</i> < 0.000001), while mean SANE improved from 54.7 to 65.3 (<i>Δ</i> =  + 10.5 points; +19.2%). The MCID was achieved by 50.2% of patients, and 74.4% showed any improvement. Improvement was observed across all anatomical locations and symptom durations. In patients with chronic symptoms lasting more than one year (<i>n</i> = 117; 57.6%), 74.4% of patients improved and 51.3% achieved the MCID (t = -9.56; <i>p</i> < 0.001), with no significant difference across chronicity groups (one-way ANOVA: F = 0.485; <i>p</i> = 0.616). Patients with higher baseline pain showed greater improvements (NRS ≥ 6: <i>Δ</i>NRS = -2.53; <i>Δ</i>SANE =  + 14.1). Satisfaction was high, with 64.0% of patients reporting satisfaction scores ≥ 4/5.</p><p><strong>Conclusions: </strong>A standardized digital intervention combining mindfulness and guided motor imagery was associated with rapid and clinically relevant improvements in pain and perceived function across all musculoskeletal regions and patient profiles, including chronic pain populations. Randomized controlled studies with objective outcomes are required to confirm these findings and investigate potential mechanisms.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1895131"},"PeriodicalIF":5.2,"publicationDate":"2026-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13635368/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119457","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial intelligence for alcohol, opioid, and cannabis use disorders screening and management: a narrative review of barriers and facilitators to clinical implementation.","authors":"Isaac Frimpong, Benedictus Atsu Hlormenu","doi":"10.3389/fdgth.2026.1958597","DOIUrl":"https://doi.org/10.3389/fdgth.2026.1958597","url":null,"abstract":"<p><p>Substance use disorder (SUD) remains one of the most prevalent and undertreated conditions in medicine. Machine learning and artificial intelligence (AI) have produced numerous predictive models for SUD risk stratification, screening, and management, but few have progressed beyond development and validation into sustained clinical implementation. This review focuses on alcohol, opioid, and cannabis use disorders, the substances for which a deployed or near-deployed AI evidence base currently exists, synthesizing barriers and facilitators to AI implementation using a hybrid framework integrating the Framework for AI Implementation Research in Healthcare (FAIIR-H) with the Unified Theory of Acceptance and Use of Technology (UTAUT) across four domains: data and model, clinician and workflow, patient, and system and regulatory factors. Alcohol use disorder has the largest predictive literature by volume but remains methodologically heterogeneous with limited external validation; opioid use disorder has a smaller but more methodologically mature and fairness-audited literature; cannabis use disorder has a more limited evidence base. Two real-world deployments illustrate this gap being bridged, with differing strength of evidence: a hospital-based opioid AI screener, supported by fairness-auditing and implementation-outcome evidence, was associated, as a secondary pre-post finding, with 47% lower odds of 30-day readmission across more than 51,000 hospitalizations; an alcohol relapse-management platform was associated with up to an 18% reduction in relapse risk within a platform dataset of more than 500,000 patient-days of observational data, though comparable implementation-outcome and fairness evidence was not identified for it. Among the studies identified in this review, no substance has yet demonstrated success in both screening and management simultaneously. Implementation is further shaped by stigma, data-sharing concerns, digital access, and the SUD-specific confidentiality requirements of 42 CFR Part 2. We propose a three-tier typology of management-focused AI applications showing that patient trust burden escalates with directness of AI-patient interaction rather than predictive accuracy and outline a measurable research and policy agenda targeting external validation, fairness auditing, and implementation in safety-net settings.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1958597"},"PeriodicalIF":5.2,"publicationDate":"2026-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13635661/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Large language models in healthcare: applications, evaluation frameworks, and governance pathways - a scoping review and multidimensional framework.","authors":"Joao C Ferreira, Isabel Rosa","doi":"10.3389/fdgth.2026.1865568","DOIUrl":"https://doi.org/10.3389/fdgth.2026.1865568","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Large language models (LLMs) are increasingly evaluated for healthcare applications spanning clinical documentation, decision support, patient communication, research assistance, and operational workflows. Despite rapid adoption interest, evidence remains heterogeneous and standardised approaches for evaluation and governance are not yet consistently applied.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;To map healthcare applications of LLMs, synthesise reported outcomes and risks, and propose a multidimensional evaluation and governance framework oriented toward digital-health implementation.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;A scoping review was conducted following the methodological guidance of Arksey and O'Malley, Levac et al., and the Joanna Briggs Institute, with reporting aligned to the PRISMA-ScR checklist. The protocol was prospectively registered on the Open Science Framework https://doi.org/10.17605/OSF.IO/SWP78). Searches were performed in PubMed/MEDLINE, Embase, Scopus, Web of Science, IEEE Xplore, ACM Digital Library, and the ACL Anthology, supplemented by medRxiv/bioRxiv preprint searches and grey-literature scanning of WHO, FDA, and EU AI-Office guidance, covering 1 January 2019-30 April 2026. Title/abstract and full-text screening were carried out in duplicate; inter-rater agreement was Cohen's κ = 0.78. Data were extracted in duplicate using a piloted form. Synthesis followed Braun and Clarke's six-phase reflexive thematic analysis. The complete list of 78 included studies is provided in Supplementary File S4.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Seventy-eight studies met inclusion criteria; 68% were published between 2023 and 2025. Five application domains were identified: (1) clinical documentation and summarisation; (2) clinical decision support and reasoning assistance; (3) patient communication and health-literacy support; (4) biomedical research and knowledge synthesis; and (5) administrative and operational use cases. The evidence base was dominated by benchmark and simulated-workflow studies (89%), with limited prospective workflow-embedded evaluations (11%). Reported benefits concentrated on documentation efficiency, text quality, and knowledge synthesis; safety-relevant risks included hallucinated content, omission of clinically critical information, demographic bias, privacy vulnerabilities, limited explainability, and automation bias. Studies were geographically concentrated in North America and East Asia, with limited representation from Sub-Saharan Africa, South Asia, and Latin America.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;Current evidence supports cautious deployment of LLMs in selected healthcare tasks under structured oversight. Translational progress depends on prospective evaluation, standardised reporting, equity-focused audits, and lifecycle governance with continuous monitoring. The proposed five-dimensional framework (technical performance, clinical validity, equity, workflow integration, gove","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1865568"},"PeriodicalIF":5.2,"publicationDate":"2026-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13632479/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From fragmentation to federation: multi-partner OMOP implementation in Uganda enabling global real-world evidence generation.","authors":"Francis Kanyike, Annet Nanungi, Harriet Dickinson, Atif Adam, James Brash, Thu Do, Caroline Otike, Michael Bogart, Alex Asiimwe, Mui Van Zandt, Cissy Kityo Mutuluuza","doi":"10.3389/fdgth.2026.1907568","DOIUrl":"https://doi.org/10.3389/fdgth.2026.1907568","url":null,"abstract":"<p><strong>Background: </strong>Incorporating geographically diverse populations into real-world evidence studies is essential for generating globally applicable insights. However, many African healthcare institutions operate highly customized electronic medical record systems that lack interoperability with international data networks. This paper describes a collaborative implementation to transform legacy electronic medical record data into the OMOP Common Data Model at the Joint Clinical Research Centre (JCRC) in Uganda, enabling participation in federated real-world evidence networks.</p><p><strong>Methods: </strong>A tri-partite collaboration was established between JCRC, IQVIA, and Gilead Sciences. The project was executed in three phases: (1) data assessment and architecture design, (2) extract, transform, and load execution and custom terminology mapping, and (3) validation of the transformed data through a cohort characterization study. JCRC's Integrated Clinical Enterprise Application (ICEA) database, containing over 48,000 patient records, was assessed and transformed to OMOP CDM v5.4. Following conversion, a descriptive observational study characterized four cohorts: all patients, people with HIV (PWH), hepatitis B virus (HBV)-infected, and hepatitis C virus (HCV)-infected patients.</p><p><strong>Results: </strong>The source database comprised 168 tables and 1,894 fields with no standardized medical coding systems -100% custom terminology- requiring manual mapping. Over 100 h of collaborative mapping achieved 95% data capture for core domains. A total of 48,748unique patients were successfully mapped to the OMOP CDM. The cohort characterization demonstrated the utility of the transformed data, revealing: HIV prevalence of 38.1% (<i>n</i> = 18,588); HBV prevalence of 7.0% (<i>n</i> = 3,423); and no HCV-infected patients identified. People with HIV demonstrated high care engagement (95.9% on antiretroviral therapy, a median of 25 visits per patient over a median follow-up of 7.6 years). Key data gaps included limited documentation of non-communicable diseases and mortality. Critical lessons included the need for detailed column-level data dictionaries and structured capacity building for local OMOP sustainability.</p><p><strong>Conclusions: </strong>This collaboration demonstrates that full OMOP CDM implementation is feasible in highly customized data environments when built on a collaborative design that combines local clinical expertise with global technical support. The project successfully positioned JCRC to engage with OHDSI's federated ecosystem and provides a replicable model for other African institutions seeking to generate representative real-world evidence at scale.</p>","PeriodicalId":73078,"journal":{"name":"Frontiers in digital health","volume":"8 ","pages":"1907568"},"PeriodicalIF":5.2,"publicationDate":"2026-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13628579/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119460","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}