BMJ digital health & AI最新文献

{"title":"When hospitals build their own AI: a comparison of the EU and US regulatory framework.","authors":"Sofia Palmieri, Timo Minssen, I Glenn Cohen","doi":"10.1136/bmjdh-2026-000105","DOIUrl":"10.1136/bmjdh-2026-000105","url":null,"abstract":"<p><p>Hospitals are increasingly shifting from being passive purchasers of artificial intelligence (AI) systems to active developers of medical AI, particularly in the form of in-hospital machine learning tools tailored to local clinical needs. This paper examines how the European Union (EU) and the USA regulate in-hospital-developed AI-based medical devices, focusing on the legal and institutional conditions that shape innovation, oversight and diffusion. Through a comparative analysis across three dimensions-responsiveness to clinical need, distribution of innovative capacity and pathways for scaling and diffusion-the paper shows that both systems enable hospital-based innovation but embed it in different regulatory logics. The EU adopts a structured model that treats hospitals as regulated innovators operating under defined institutional obligations, whereas the USA relies on a more permissive and less institutionally differentiated framework in which hospital innovation emerges as a by-product of general software regulation.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000105"},"PeriodicalIF":0.0,"publicationDate":"2026-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13599951/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149043417","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 Long road to AI-physician partnership starts with learners: a map for AI in medical education.","authors":"Nikhil Patel, Andre Kumar, Jeffrey Chi, Kevin Keet, John Kugler, Jason Hom, Poonam Hosamani","doi":"10.1136/bmjdh-2026-000056","DOIUrl":"10.1136/bmjdh-2026-000056","url":null,"abstract":"<p><p>Medical students and residents are already using generative artificial intelligence (AI) to draft notes, summarise records and generate differential diagnoses, often informally and without institutional oversight. Adoption has outpaced policy and pedagogical processes, which makes restriction an unrealistic response. The practical question is how to integrate these tools so that clinical skills are not replaced as they are being built. We aim to give medical educators a practical guide for judging which uses of generative AI support skill formation at each stage of training and which displace it. Three distinct harms are at stake: deskilling, the erosion of acquired competencies; never-skilling, the failure to develop foundational skills when AI performs them before learners do; and mis-skilling, the absorption of AI-generated errors and biases as accurate practice. To calibrate integration, we propose a framework built on cognitive displacement, the degree to which a tool takes over the reasoning that training is designed to build. Appropriate use depends on a task's position on that spectrum and the learner's stage. Applied across the training continuum, the framework locates each risk where it has the greatest potential for harm. Never-skilling threatens preclinical students who have not yet built the competencies that AI could erode. Deskilling threatens residents, whose judgement is still consolidating. Mis-skilling arises wherever biased tools meet trainees making unsupervised decisions. Because the framework's stage-specific predictions remain largely untested, we set out an agenda for measuring these risks prospectively. The principle is calibration, not prohibition. The open question is who designs that integration, and when.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000056"},"PeriodicalIF":0.0,"publicationDate":"2026-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13583697/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148993688","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":"Beyond productivity: a <i>NASSS-informed review of implementation risks and research priorities for ambient AI scribes in healthcare</i>.","authors":"Lucas Martinus Seuren, Robin Williams, Kathrin Cresswell","doi":"10.1136/bmjdh-2026-000091","DOIUrl":"https://doi.org/10.1136/bmjdh-2026-000091","url":null,"abstract":"<p><p>Ambient artificial intelligence scribes (ambient scribes) are increasingly used across healthcare systems to support clinical documentation. Implementation strategies and associated research have been mainly concerned with the ability to accurately and efficiently produce clinical notes. Some concerns have been raised, such as potential hallucinations or omissions that require editorial work of clinicians, but this research has not moved beyond the technical performance of ambient scribes. We sought to synthesise emerging evidence and perspectives on ambient scribes from a sociotechnical perspective, focusing on how these systems interact with existing clinical work practices and organisational processes. We searched Embase, PubMed and Scopus to identify relevant literature and included 27 articles. We used a sociotechnical lens to synthesise research and map core themes against implementation domains from the Non-Adoption, Abandonment, Scale-Up, Spread and Sustainability (NASSS) framework. We identified few empirical studies examining the risks of ambient scribes, and even fewer investigating their implementation and adoption in practice. Existing research has focused predominantly on their potential to reduce administrative burden and improve efficiency. However, ambient scribes do more than automate note-taking. They actively shape clinical workflows and patient-clinician interactions and change the nature and function of clinical documentation. This suggests that there has been a de facto acceptance by policymakers and practitioners of the value proposition without critically examining the underlying assumptions and broader consequences. Ambient scribes must be understood as complex interventions rather than just productivity tools. There is therefore an urgent need for contextual validation and longitudinal evaluation.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000091"},"PeriodicalIF":0.0,"publicationDate":"2026-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13663716/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149198848","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":"Bridging the digital pathology divide: reciprocal learning between high-resource and low-resource settings.","authors":"Kamran M Mirza, Arvind Rao, Jessica A Baker, Robert K Parker, Shane C Quinonez, Priscilla Njenga, Mansoor Saleh, Ulysses Gj Balis, Akbar K Waljee, Shahin Sayed","doi":"10.1136/bmjdh-2026-000026","DOIUrl":"10.1136/bmjdh-2026-000026","url":null,"abstract":"<p><p>Digital pathology (DP) represents a major advance in global health, integrating digital whole-slide imaging, telepathology and clinical-grade artificial intelligence into cohesive workflows that can improve diagnostic accuracy and educational access. Realising these gains, however, depends on local infrastructure, governance, reimbursement and workforce capacity. The COVID-19 pandemic accelerated DP uptake in some settings, but adoption remains highly uneven between high-resource settings (HRS) and low-resource settings (LRS), reflecting unequal infrastructure, workforce, policy and financing. This analysis contends that reciprocal innovation, a model of ongoing two-way learning between HRS and LRS that is distinct from one-way technology transfer, can accelerate DP adoption by leveraging the strengths of each environment. Drawing on examples from HRS and LRS, we examine use cases in education and clinical practice, the adoption barriers that persist and policy mechanisms required for sustainable scaling. While HRS contribute regulatory frameworks, validation standards and mentorship, LRS offer innovations in cost efficiency, workflow resilience and open-source adaptation that can inform more sustainable models globally. Neither HRS nor LRS are homogeneous entities. Resource availability also varies substantially within each category: HRS may contain poorly resourced institutions, while some institutions in LRS may have comparatively greater resources but still face broader system-level infrastructure constraints. Ensuring access to those in LRS requires targeted investment, harmonised standards and mutual learning across resource levels. DP has the potential to foster bidirectional partnerships to democratise access to timely, accurate and patient-centred pathology services across all resource settings.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000026"},"PeriodicalIF":0.0,"publicationDate":"2026-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13548248/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148915111","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":"Staged, life-cycle approach to evidence generation for digital health interventions in low- and middle-income countries.","authors":"Gareth Obery, Eva Weicken, Shubhanan Upadhyay, Max Rath, Bilal A Mateen, Saira Ghafur","doi":"10.1136/bmjdh-2026-000059","DOIUrl":"10.1136/bmjdh-2026-000059","url":null,"abstract":"<p><p>Digital health interventions (DHIs) are rapidly expanding across low- and middle-income countries (LMICs), driven by their potential to address persistent health system constraints, workforce shortages and health inequities. However, the evaluation of these interventions has not kept pace with their adoption. Traditional evidence-generation approaches, particularly those centred on rigid, resource-intensive study designs, are often poorly aligned with the iterative nature of digital technologies and the realities of resource-constrained settings. This misalignment can exacerbate existing system pressures, slow validation and contribute to an enduring evidence-implementation gap that limits sustainable adoption and scale. This narrative review synthesises peer-reviewed biomedical literature, grey literature, institutional reports and applied case studies published between 2015 and 2025 to examine how evidence for DHIs is generated in LMICs. The analysis is organised across three interrelated domains: contextual and stakeholder dynamics shaping evaluations, methodological innovations suited to LMIC realities and top-down policy and regulatory mechanisms influencing adoption and scale. Drawing on this synthesis, the review argues for a staged, life-cycle-based approach to evidence generation, in which evaluation methods and evidence strategies evolve as the intervention matures and the implementation context evolves. This approach emphasises pragmatic, context-sensitive strategies that balance scientific rigour with feasibility. This review aims to provide decision-oriented guidance for innovators, implementers, policy-makers and regulators seeking to generate appropriate, actionable and policy-relevant evidence to support the adoption, integration and scale of DHIs in LMIC health systems.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000059"},"PeriodicalIF":0.0,"publicationDate":"2026-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13548215/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148915140","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":"Prognostication and clinical opportunities with AI for coronary artery calcium: a scoping review.","authors":"Valerie J F McDonald, Shyon Parsa, Domenico Mastrodicasa, Fatima Rodriguez, Curtis Langlotz, David J Maron, Alexander T Sandhu, Ramzi Dudum, Sneha S Jain","doi":"10.1136/bmjdh-2026-000053","DOIUrl":"10.1136/bmjdh-2026-000053","url":null,"abstract":"<p><strong>Objective: </strong>Coronary artery calcium (CAC) scoring is an established marker of atherosclerotic burden and cardiovascular risk, yet its clinical use remains limited by access, cost and workflow constraints. Artificial intelligence-enabled CAC (AI-CAC) tools allow automated assessment of coronary calcification from chest CT performed for diverse clinical indications, creating opportunities for scalable, opportunistic cardiovascular disease risk stratification.We conducted a scoping review of AI-CAC models focused on prognostication and real-world deployment, emphasising outcomes, clinical value and implementation considerations.</p><p><strong>Methods and analysis: </strong>Peer-reviewed original studies using an AI model incorporating CAC from chest CT for clinical prognostication, risk prediction or assessment of clinical interventions, reporting clinical outcomes, prognostic metrics or implementation insights. Studies focused exclusively on technical AI-CAC development or validation without clinical context were excluded.PubMed/MEDLINE, Embase and Cochrane Library, searched through November 2025, supplemented by expert recommendations from study coauthors.A single reviewer screened all records. Data were extracted using a standardised 25-variable template and synthesised narratively using an AI translational maturity framework.</p><p><strong>Results: </strong>Across diverse patient populations and imaging contexts, AI-CAC demonstrated consistent associations with all-cause mortality, cardiovascular mortality, major adverse cardiovascular events and obstructive coronary artery disease. Radiomics-based approaches incorporating high-dimensional imaging features may improve prognostic performance beyond Agatston scoring alone, demonstrating further methodological potential for assessing cardiovascular risk. Emerging prospective and implementation-focused studies suggest that integrating AI-CAC into clinical workflows increases preventive care engagement. However, most evidence remains retrospective with limited evaluation of cost-effectiveness, equity or long-term patient and health system impact.</p><p><strong>Conclusion: </strong>AI-CAC represents a promising approach for scaling atherosclerotic cardiovascular disease risk stratification. Beyond expanding access to traditional CAC scoring, AI-enabled approaches may support the evolution of the CAC score itself by integrating plaque characteristics, distribution and complementary imaging biomarkers to support more personalised preventive strategies rather than relying on fixed risk thresholds alone. Realising meaningful clinical impact will require prospective implementation studies, standardised reporting and evaluation, integration into care pathways and ongoing monitoring aligned with ethical, operational and financial considerations.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000053"},"PeriodicalIF":0.0,"publicationDate":"2026-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13536142/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148915127","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":"Beyond protected characteristics: rethinking AI fairness through the lens of structural and social determinants of health.","authors":"Karol Basta, Uzma Zahid, Mahi Hardalupas, Dijana Vilic","doi":"10.1136/bmjdh-2026-000087","DOIUrl":"10.1136/bmjdh-2026-000087","url":null,"abstract":"","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000087"},"PeriodicalIF":0.0,"publicationDate":"2026-07-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528398/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148915098","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":"How artificial intelligence is disrupting the field of medical knowledge and evaluation: a narrative review.","authors":"Kathrin Cresswell, Robin Williams","doi":"10.1136/bmjdh-2026-000104","DOIUrl":"10.1136/bmjdh-2026-000104","url":null,"abstract":"<p><p>Although artificial intelligence (AI) shows considerable promise in healthcare, its introduction and safe and effective exploitation involve several challenges. We argue that to maximise the benefits of AI and understand, track and mitigate potential harms there must be a shift in how AI is evaluated. Evaluation should be more closely linked to the local contexts in which AI is deployed and used and better integrated with clinical practice. To support this shift, we propose three interrelated levels of evaluation: evaluation of implementation, which examines transformations in contextually situated practices; evaluation of optimisation, which explores changes to clinical pathways over extended periods; and evaluation of scaling, which focuses on coordination through regional and national structures. This approach offers significant opportunities to advance both healthcare practice and the epistemology of medical knowledge.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000104"},"PeriodicalIF":0.0,"publicationDate":"2026-07-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528392/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148915025","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 use of ambient voice technology (AVT) for clinician-patient consultations in healthcare practice: A rapid systematic scoping review.","authors":"Rachel Lawrence, Kevin Herbert, Theo Georghiou, Raj Mehta, Pei Li Ng, Lucina Rolewicz, Stephen Morris, Jenny Shand","doi":"10.1136/bmjdh-2026-000120","DOIUrl":"10.1136/bmjdh-2026-000120","url":null,"abstract":"<p><strong>Objective: </strong>Using ambient voice technology (AVT) to transcribe and summarise clinician-patient consultations in real time is increasing in healthcare practice, with the aim of reducing clinical documentation burden. However, the evidence base remains underdeveloped. This review synthesised current evidence on the use and impact of AVT for documenting clinician-patient consultations in healthcare. We also examined how impact has been evaluated and measured with the aim of informing future research and assessment in real-world practice.</p><p><strong>Methods: </strong>We conducted a rapid systematic scoping review (PROSPERO: CRD420251155579). Four databases and grey literature sources were searched in September 2025. Of 598 articles screened, 21 met the inclusion criteria. These studies evaluated user experiences with AVT (n<i>=</i>13), quantified impacts using objective numerical data (n<i>=</i>19) and perceived impacts based on self-reported experiences (n<i>=</i>17).</p><p><strong>Results: </strong>Evidence on the real-world impact of AVT remains limited. While some studies report reduced documentation burden and improvements in staff well-being and perceived care quality, these findings largely stem from short-term observational designs and self-reported measures. Evidence relating to wider, system-level impacts such as pathway productivity, patient safety and cost effectiveness is sparse. Approaches to evaluating AVT were heterogeneous and inconsistent, particularly when reporting on system-level impacts beyond the clinician-patient consultation. Different factors (e.g., workflow integration, output quality), the consultation context and clinical setting influenced experiences of using AVT.</p><p><strong>Conclusions: </strong>Current evidence does not yet support widespread adoption of AVT in clinical practice. Although early findings suggest potentially important benefits, available studies are heterogeneous, predominantly short-term and methodologically limited. It remains unclear whether the improvements reported to date translate into sustained gains in clinical capacity, care quality or system efficiency. Longer-term evaluations using standardised outcome measures complemented by qualitative methods, are needed to determine the impact and value of AVT in healthcare.</p><p><strong>Prospero registration number: </strong>CRD420251155579.</p>","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000120"},"PeriodicalIF":0.0,"publicationDate":"2026-07-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528325/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148915139","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":"Generative AI for mental health: promise, risk and the path forward.","authors":"Matthew Brown, Naomi Herz","doi":"10.1136/bmjdh-2026-000158","DOIUrl":"10.1136/bmjdh-2026-000158","url":null,"abstract":"","PeriodicalId":524116,"journal":{"name":"BMJ digital health & AI","volume":"2 1","pages":"e000158"},"PeriodicalIF":0.0,"publicationDate":"2026-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528334/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148915062","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}