bims-aimedu Biomed News
on AI in medical education
Issue of 2026–03–15
twelve papers selected by
Angela Spencer, Saint Louis University



  1. JMIR Form Res. 2026 Mar 09. 10 e76838
       BACKGROUND: Artificial intelligence (AI) is increasingly influencing medical student education, with AI-driven chatbots, such as ChatGPT, emerging as powerful study tools. While these technologies offer numerous benefits, they also pose challenges that warrant the adaptation of medical school curricula.
    OBJECTIVE: This study examines medical students' perceptions and use of ChatGPT. We hypothesize that ChatGPT is widely used for academic support, but concerns remain regarding reliability and academic integrity.
    METHODS: We conducted a cross-sectional study from August 25 to December 10, 2024, in the United States. Students in all years of medical training who were enrolled in accredited allopathic or osteopathic medical schools were eligible to participate. Data were collected using an anonymous online questionnaire, which was distributed through institutional mailing lists. Overall, 188 schools were reached, of which 14 (7.4%) responded and agreed to distribute the survey. A total of 177 participants completed the survey. Survey items consisted primarily of Likert-scale and multiple-choice questions. Primary outcome measures included self-reported frequency of ChatGPT use, perceived usefulness of ChatGPT, and ChatGPT use habits.
    RESULTS: Overall, 98.9% (175/177) of participants had heard of ChatGPT, with 88.7% (157/177) reporting having used it; 62.7% (111/177) identified as female, and 52% (92/177) had completed at least 1 block of clinical rotations. Medical students most often used ChatGPT to understand complex medical concepts, prepare for exams, and generate study materials. Moreover, 46.5% (73/157) used it to help complete medical school assignments. Medical students also reported using it clinically, with the most common use being to generate differential diagnoses. Notably, 21.0% (33/157) of participants responded having used ChatGPT to help write clinical notes. Moreover, 73.9% (116/157) reported that their experience with ChatGPT improved their overall perception of AI's potential to assist in medical practice, and 86.6% (135/157) believed that having ChatGPT as a resource would make them more effective physicians. Statistical analyses were performed using the Pearson chi-square test with α=.05. Students who reported moderate or advanced baseline understanding of AI were more likely to practice conscientious use habits, such as cross-checking (odds ratio [OR] 2.31, 95% CI 1.08-4.97) and editing (OR 2.45, 95% CI 1.05-5.71) ChatGPT output before using it, than those who reported a basic or limited understanding.
    CONCLUSIONS: Our study is among the few to examine medical student perceptions of ChatGPT at a national level. We examined responsible use habits to identify areas in which reliance on this technology may lead users astray. We found that ChatGPT is being used to complete academic assignments and write clinical notes, raising concerns about information verification, AI literacy, patient confidentiality, and ethical use. Together, these findings highlight the need for structured AI education to help students leverage these technologies effectively while mitigating risks associated with misinformation and overreliance on AI.
    Keywords:  AI; ChatGPT; artificial intelligence; large language model; medical education; medical student
    DOI:  https://doi.org/10.2196/76838
  2. JMIR Med Educ. 2026 Mar 03. 12 e86686
       Unlabelled: As artificial intelligence (AI) develops, the medical education community has begun defining the relevant forms of competency. Many experts emphasize the importance of optimizing AI tools' output or understanding the relevant technical and normative considerations around using AI tools. A recent publication in this journal showed that optimizing instructions for large language models may yield diminishing returns as such tools improve. This suggests the need for a new competency-one that focuses on choosing the appropriate AI tools. I briefly summarize the current competency domains and examples to contextualize the current state of AI competency development, highlighting the need for further synthesis. I then introduce a hierarchical framework of competencies that might assist with priority-setting around subsequent competency development work. It consists of cognitive, operational, and meta-AI domains, which respectively correspond with the knowledge around understanding, using, and choosing AI tools. The final section describes the potential challenges associated with the development of AI competency. These include traditional concerns around competency-based medical education: deciding whether and which competencies are meaningful for measuring the targets of interest; adjusting the relevant measurements to reflect the necessary temporal and institutional context; and setting up the relevant organizational support to encourage measurement of competency. This section also discusses the challenges of developing the relevant performance indicators for AI tools across different clinical contexts. Such indicators will be necessary for guiding the choice of AI tools for the clinical context, but medical educators may not have the skills to develop them. In addition to identifying potential sources for relevant indicators, the medical education community may shape physicians' norms of practice to drive the AI industry into producing the relevant indicators. The potential for physicians to incur higher medical liability from poor choice of AI may lead them to demand more nuanced performance indicators from AI suppliers. Physicians are also in a position to do so, since the competitive AI market may provide them more bargaining power.
    Keywords:  AI; artificial intelligence; medical education
    DOI:  https://doi.org/10.2196/86686
  3. Med Sci Educ. 2025 Dec;35(6): 2729-2733
      Artificial intelligence (AI) imitates human intelligence using computer systems to carry out tasks traditionally done by humans. Self-efficacy refers to the individual's confidence in carrying out tasks or achieving set goals. Clinical education refers to the process of learning and teaching in the area of diagnosis, prevention and treatment of diseases. In clinical education, various AI tools are used for teaching and learning, and are said to help improve self-efficacy. This article delineated how the use of AI in clinical education potentially stimulates Bandura's sources of self-efficacy, mastery experience, vicarious experience, verbal or social persuasion and physiological and affective states to enhance learning and teaching experiences.
    Keywords:  Artificial intelligence; Clinical education; Learning; Self-efficacy; Teaching
    DOI:  https://doi.org/10.1007/s40670-025-02572-9
  4. Med Sci Educ. 2025 Dec;35(6): 2751-2762
       Background: Generative AI (GenAI) presents both opportunities and challenges for higher education. While it offers the potential to personalise learning and improve educator processes, concerns around academic integrity and output accuracy persist. Health professionals must navigate this landscape carefully to ensure technology augments, rather than compromises, the development of core clinical and professional competencies in higher education.
    Objective: This study aimed to develop a framework for implementing GenAI into the curriculum. To achieve this, a synthesis of the existing evidence on the applications, benefits, and challenges of GenAI in health professions education was required.
    Methods: To achieve the required knowledge for the development of the framework, a systematic scoping review was conducted following the Joanna Briggs Institute (JBI) methodology and reported using the PRISMA-ScR guidelines. A comprehensive search of MEDLINE, CINAHL, Scopus, Web of Science, and ERIC databases was performed to identify relevant studies. A narrative synthesis was used to map the literature across key themes to inform the framework's development. Results: The review included 21 studies, which highlighted the use of GenAI by educators and students to aid in productivity and learning. Key challenges identified included the risk of generating inaccurate content, the potential for misuse, and the critical need for enhanced GenAI literacy among both students and staff. The findings were synthesised into three primary domains: educator use, student use, and assessment design and purpose.
    Conclusion: The integration of GenAI in health education requires a structured, proactive approach. We propose an evidence-informed framework centred on three core pillars: Student GenAI Literacy, Educator Capability, and Assessment Design. This framework provides a roadmap for institutions to harness GenAI responsibly, ensuring it serves as a tool to support critical thinking and professional judgement.
    Supplementary Information: The online version contains supplementary material available at 10.1007/s40670-025-02578-3.
    Keywords:  AI Literacy; Curriculum Development; Generative AI; Health Professions Education; Medical Education; Scoping Review
    DOI:  https://doi.org/10.1007/s40670-025-02578-3
  5. Qual Health Res. 2026 Mar 07. 10497323261425889
      Artificial intelligence (AI) technologies are rapidly expanding in qualitative health research and often promise improved efficiency or novel discoveries. However, this promise has yet to be realized, and further, serious ethical issues emerge ranging from the use of AI videoconferencing technologies to conduct interviews, AI transcription services, and AI-augmented qualitative analysis tools. These ethical dilemmas are not always obvious and require careful consideration of the ramifications of integrating these technologies in the research process. These concerns are relevant to all stages of research experience ranging from emerging scholars to more practiced researchers but are particularly significant in training new scholars who are early adopters of AI technologies. To trace the ethical issues surrounding AI in the practice of qualitative health research, we map the specific values of autonomy, privacy, validity, and equity to highlight decision points and provide a framework for navigating ethical use of the AI tools.
    Keywords:  artificial intelligence; ethics; qualitative research; technology
    DOI:  https://doi.org/10.1177/10497323261425889
  6. J Nurs Scholarsh. 2026 Mar;58(2): e70076
       INTRODUCTION: Systematic reviews (SRs) require comprehensive, reproducible searches, yet developing search strategies is resource-intensive and demands specialized expertise. Generative AI offers potential to streamline this process, but empirical evaluations for GAI-assisted SR searching remain scarce. The objectives of this study are to: demonstrate a step-by-step process for developing a custom ChatGPT-based chatbot to support SR search strategy development, and evaluate its performance.
    DESIGN: A cross-sectional evaluation study.
    METHODS: We used ChatGPT-4.0 to create a chatbot designed to mimic a medical librarian, generating PICO-informed searches. Its knowledge base was augmented with two methodological references. After piloting testing, we refined its instructions. For evaluation, we randomly sampled 50 Cochrane SRs published in 2024. Standardized P-I-O prompts produced database-ready queries for PUBMED and EMBASE. The primary outcome was per-review success rate, summarized by median and inter-quartile range. A sensitivity analysis was conducted.
    RESULTS: Pilot testing achieved a retrieval rate of 41/49 (83.7%). In the main sample (1169 studies; median 13.5 studies per SR), the chatbot identified a median of 67.4% of included studies (IQR: 43.1%-88.4%). When limited to indexed studies (n = 1114), retrieval rose to 72.0% (IQR: 46.0%-92.5%). Lower performance was observed when outcomes were absent from the abstracts or interventions had many lexical variants.
    CONCLUSIONS: A GAI-based chatbot can rapidly generate SR searches (~67%-72% identification), serving as a useful starting point but not a replacement for expert-led approaches. Integration of librarian expertise, structured prompts, and controlled vocabularies may improve performance. Further benchmarking and transparent reporting are needed to guide adoption.
    Keywords:  database searching; generative artificial intelligence; large language model; systematic review
    DOI:  https://doi.org/10.1111/jnu.70076
  7. East Afr Health Res J. 2025 ;9(1): 9-19
       Background: Artificial Intelligence (AI) is transforming medical education by enabling personalised learning, adaptive feedback, simulation-based training, and automated assessments. While AI offers significant benefits, including curriculum optimisation and virtual tutoring, concerns around data privacy, access, and ethical implementation persist. Although bibliometric studies have explored AI in healthcare, comprehensive analyses of global collaboration and publication trends in AI-focused medical education remain limited.
    Aim: This study aims to analyse global research trends, key contributors, and thematic developments in the application of AI within medical education.
    Methods: A bibliometric analysis was conducted using the Scopus database. The search strategy included terms such as "Medical Education", "Artificial Intelligence", "Machine Learning", "Deep Learning", "Clinical Training", "Virtual Patients", and "Simulation". Data were analysed using the Bibliometrix R package to assess publication volume, keyword co-occurrence, author collaboration, and citation patterns.
    Results: Research output on AI in medical education has grown significantly, peaking in 2024 with 1,081 publications. The United States leads in publication volume, followed by Russia and Canada. "Artificial Intelligence" was the most frequently used keyword. Co-authorship and co-citation networks revealed strong international collaboration, with emerging themes in clinical competence, virtual simulation, and ethical considerations.
    Conclusion: The future of artificial intelligence in medical education is promising, with applications in personalised treatment plans, drug development, and virtual healthcare assistants. AI has transformative potential for medical education, particularly in personalised learning and simulation-based training. Strategic investment in AI literacy, ethical frameworks, and infrastructure is essential to ensure equitable and effective integration across global contexts.
    DOI:  https://doi.org/10.24248/eahrj.v9i1.817
  8. BMJ Open. 2026 Mar 10. 16(3): e099887
       OBJECTIVES: To examine which information sources medical specialists use to answer clinical questions in daily practice and to describe the relative frequency of use for each source.
    DESIGN: Systematic review with narrative synthesis and meta-analysis.
    DATA SOURCES: Academic Search Premier, APA PsycINFO, CINAHL, Emcare, Cochrane Library, Web of Science, Embase and PubMed were searched for relevant studies published from 2000 to 1 June 2025.
    ELIGIBILITY CRITERIA: We included peer-reviewed English-language studies reporting on the frequency of information source usage by medical specialists when addressing clinical questions. Studies reporting usage on a continuous (0-100%) scale were eligible for meta-analysis.
    DATA EXTRACTION AND SYNTHESIS: Two reviewers independently screened studies. Data were extracted by one reviewer and checked by a second. Study quality was assessed using the Quality Assessment tool with Diverse Studies tool. A narrative synthesis was conducted for studies that were not eligible for quantitative pooling to summarise patterns in information-seeking behaviour and reported barriers. A random-effects meta-analysis was performed for studies reporting continuous usage percentages and assessing at least four information sources. Sensitivity analyses were conducted using a leave-one-out approach. Potential publication bias was explored descriptively using funnel plots.
    RESULTS: 25 studies were included, of which 6 (with 8641 participants) were eligible for meta-analysis. The narrative synthesis of non-pooled studies showed a consistent reliance on standalone information sources and identified barriers to the use of aggregated sources. In the meta-analysis, digital databases such as PubMed were the most frequently used information source (74%, 95% CI 63% to 85%), followed by textbooks (71%, 95% CI 57% to 85%) and consultation with colleagues (43%, 95% CI 15% to 71%). Systematically aggregated sources, including clinical practice guidelines (38%, 95% CI 27% to 49%) and point-of-care websites (49%, 95% CI 17% to 81%), were used less frequently. Sensitivity analyses indicated that pooled estimates were generally robust, although results should be interpreted cautiously given methodological variability across studies.
    CONCLUSIONS: Medical specialists predominantly rely on standalone information sources when addressing clinical questions, while systematically aggregated and interpreted sources such as clinical practice guidelines and point-of-care tools are used less frequently. These findings highlight the need to better understand and address barriers to the use of aggregated information sources in clinical practice.
    PROSPERO REGISTRATION NUMBER: CRD42022267431.
    Keywords:  Evidence-Based Medicine; Information source management; clinical information; digital resources; medical specialists; systematic review
    DOI:  https://doi.org/10.1136/bmjopen-2025-099887
  9. J Med Internet Res. 2026 Mar 02. 28 e79863
    LEAPfROG Consortium
       Background: Electronic health record (EHR) data, a key form of routinely collected patient data, offer great potential for medical research and the development of artificial intelligence (AI) tools. However, because these data are primarily gathered for health care rather than research, it often lacks the quality needed for AI training, raising both methodological and ethical concerns. While previous studies have reviewed the ethical implications of both routinely collected patient data and AI separately, their intersection, where AI is applied to such data, remains largely unexplored.
    Objective: This study aimed to examine the ethical challenges that arise at the intersection of EHR data and AI development and to derive practice-oriented recommendations using the Dutch LEAPfROG (Leveraging Real-World Data to Optimize Pharmacotherapy Outcomes in Multimorbid Patients Using Machine Learning and Knowledge Representation Methods) project as a guiding case.
    Methods: We used a mixed methods design combining a scoping literature review with a systematic search and 2 stakeholder workshops structured by the guidance ethics approach, reflecting a staged and iterative process aligned with the LEAPfROG project's development phases. The review identified 25 relevant publications from 2014 to 2024. The workshops, conducted with 17 and 13 participants respectively, included patients, clinicians, ethicists, data officers, and AI developers. Both workshops used dialogue to identify ethical values, impacts, and action points, focusing on a case study of drug-induced acute kidney injury.
    Results: The analysis highlighted four major themes: (1) data privacy, transparency, and consent, including challenges of meaningful consent and risks of reidentification; (2) public trust and regulatory challenges, such as fragmented oversight and inconsistent governance; (3) fair representation and model generalizability, where incomplete or biased data may perpetuate health inequities; and (4) responsible AI integration in clinical practice, including concerns about clinical tropism, administrative burden, and the risk of overreliance on AI outputs. Both literature and stakeholder perspectives underscore the risk of decontextualization when EHR data are reused and emphasize the importance of clearly defining the purpose of data reuse to ensure real-world applicability and foster trust.
    Conclusions: Responsible AI development requires explicit attention to how EHR data are produced, interpreted, and governed in practice, recognizing that data quality and meaning are shaped by the clinical, institutional, and social contexts in which they originate. Technical solutions or top-down regulation alone are insufficient. Instead, stakeholder-led and context-sensitive approaches are needed to define the purposes, risks, and benefits of medical AI. Grounded in the realities of health care practice and in the perspectives of patients, clinicians, and data custodians, these approaches can strengthen transparency, fairness, and clinical relevance, ensuring that EHR data are used ethically and effectively to support equitable and trustworthy AI innovation.
    Keywords:  artificial intelligence; ethics; medical informatics; pharmacotherapy; routinely collected health data; stakeholder participation
    DOI:  https://doi.org/10.2196/79863
  10. J Med Educ Curric Dev. 2026 Jan-Dec;13:13 23821205261422892
       Background: Peer-assisted learning (PAL) has been increasingly adopted in undergraduate medical education as a strategy to enhance the acquisition of clinical skills. Its potential to promote collaborative learning, cost-effectiveness, and professional development makes it an attractive pedagogical approach. However, the evidence surrounding its implementation and outcomes in clinical skills teaching remains scattered.
    Methods: We conducted a scoping review following the PRISMA-ScR guidelines to map the extent, range, and nature of the literature on PAL in undergraduate clinical skills education. Relevant studies were identified through searches of PubMed, Scopus, Web of Science, and ERIC from inception to 17 September 2024. Articles were included if they examined PAL interventions involving undergraduate medical students in clinical skills training. Data were extracted on study characteristics, implementation strategies, outcomes, and reported challenges.
    Results: A total of 95 studies met the inclusion criteria. PAL was applied across a range of clinical skills including history taking, physical examination, procedural skills, and communication training. Reported benefits included improved student confidence, enhanced skill acquisition, and positive perceptions of peer tutors and learners. Challenges included variability in tutor training, assessment methods, and sustainability of programs. Evidence gaps were noted in long-term outcomes and standardization of evaluation tools.
    Conclusions: PAL is a valuable educational approach for teaching clinical skills in undergraduate medical education, with benefits for both tutors and learners. Further research is needed to establish standardized frameworks, assess long-term impact, and guide integration into formal curricula.
    Keywords:  Peer-assisted learning; clinical skills; medical curriculum; scoping review; undergraduate medical education
    DOI:  https://doi.org/10.1177/23821205261422892