Date Submitted: Jan 18, 2026. Open Peer Review Period: Jan 19, 2026 - Mar 16, 2026.

Background: Physician maldistribution remains a global challenge, with Japan’s rural regions facing critical health care shortages. Regional quota programs aim to attract medical students to underserved areas; however, their effectiveness in fostering long-term commitment is uncertain. Community-oriented medical education #mededu (COME) programs aim to address this issue by developing students’ understanding and dedication to rural health care. Objective: This study investigated the impact of an enhanced COME program, featuring increased early clinical exposure and faculty development, on first-year regional quota medical students’ perception of community health care at Chiba University. Methods: We conducted a cross-sectional qualitative study comparing 2 cohorts, 20 students enrolled from the existing COME course (April-December 2021) and 20 from the revised course (April-December 2022). The revised course included an additional day of community-based clinical exposure supervised by COME-trained attending physicians. Students’ written reflections were analyzed using qualitative content analysis and categorized according to the Fink Taxonomy of significant learning, comprising 6 domains, including foundational knowledge, application, integration, human dimension, caring, and learning how to learn. Reflections were synthesized into higher-order themes crosswalked to the Fink domains. Results: Demographics were similar between the 2021 and 2022 cohorts. In 2021, 311 learning codes were identified across foundational knowledge (n=128), application (n=91), integration (n=40), human dimension (n=16), caring (n=30), and learning how to learn (n=6). In 2022, codes increased to 385, with notable growth in caring (n=58) and human dimension (n=57), alongside increases in learning how to learn (n=15) and integration (n=45). Theme-based synthesis identified four overarching themes: (1) community health care as an interconnected, resource-constrained system; (2) patient-centered relationships and trust through communication and teamwork; (3) emerging professional identity and responsibility toward community service; and (4) developing a self-directed learning orientation for community practice. Qualitative analysis revealed that students gained a deeper understanding of patient-centered care, interprofessional collaboration, and social challenges in rural health care. The consistency in the foundational knowledge domain underscored a stable conceptual foundation, while the increase in affective and reflective domains reflected greater emphasis on interpersonal, value-oriented, and reflective learning in the revised cohort. Conclusions: Enhancements of the COME program, including additional early clinical exposure and faculty development, were associated with improved students’ perceptions of community health care. The increased focus on the caring and human dimension domains underscores the role of practical experiences in fostering collaboration, communication, and patient-centered care. The theme-based synthesis further suggests that the revised program prompted more frequent reflections on professional identity formation and self-directed learning while maintaining a stable foundation of community health care concepts. Mentorship by community hospital attendings, alongside structured clinical exposure, appears crucial in shaping medical students’ understanding and commitment to rural medicine. Ongoing longitudinal evaluations are warranted to assess the sustained impact of COME programs on career trajectories in underserved areas.

Background: Traditional Chinese medicine (TCM) has been widely used against various diseases in China for thousands of years and showed satisfactory effectiveness. However, many surveys found that TCM receives less recognition from Western medicine (WM) doctors and students. Presently, TCM is offered as a compulsory course for WM students in Western medical schools. Objective: This study aimed to investigate whether TCM courses can affect the WM students’ attitude toward TCM. Methods: WM students from Xiangya medical school were invited to finish the online questionnaire before and immediately after the TCM course. Their attitude toward TCM and treatment preferences for different kinds of diseases were tested. The Attitude Scale of TCM (ASTCM) was used. The main part of the ASTCM was designed to measure the attitude of medical students towards traditional Chinese medicine. It was composed of 18 items, divided into cognitive dimension (5 terms), emotional dimension (8 terms) and behavior tendency factor (5 terms). Results: Finally, the results of 118 five-year programs (FYP) and 36 eight-year programs (EYP) students were included. For FYP students, there was a significant increase in the total score (66.42 vs 71.43, P＜0.01) of ASTCM after the TCM course. The significant increase also showed in scores of the three factors of attitude (cognition: 21.64 vs 22.90, affection: 25.21 vs 27.96, behavior tendency: 19.57 vs 20.58, P＜0.01). Except for score of behavior tendency (17.50 vs 18.78, P＜0.05), a significant increase was not detected in total score, cognition, and affection in EPY students (total score: 60.36 vs 62.92, cognition: 20.50 vs 20.69, affection: 22.36 vs 23.44, P＞0.05). The treatment preference of FYP students in acute (P＜0.05), chronic (P＜0.05), and physical diseases (P＜0.05) showed remarkable change. The great change also was detected in internal diseases (P＜0.05), surgical diseases (perioperative period) (P＜0.05), and mental illnesses (P＜0.05) in EYP students. This change mainly appeared as a decline in WM preference and an increase in TCM & WM preference. Conclusions: The study showed that earlier TCM course offering increased the positive attitude toward TCM in students majoring in WM. The results will provide some suggestions for TCM course arrangement in WM schools. Clinical Trial: This study was approved by the Ethics Committee of Hunan First Normal University (No. 202202. The informed consent was acquired from all participants by choosing a‘yes’at the beginning of the online questionnaire.

This paper proposes a framework for leveraging large language models (LLMs) to generate misconceptions as a tool for collaborative learning in healthcare education. While misconceptions—particularly those generated by AI—are often viewed as detrimental to learning, we present an alternative perspective: that LLM-generated misconceptions, when addressed through structured peer discussion, can promote conceptual change and critical thinking. The paper outlines use cases across healthcare disciplines, including both clinical and basic science contexts, and a practical 10-step guidance for educators to implement the framework. It also highlights the need for medium- to long-term research to evaluate the impact of LLM-supported learning on student outcomes. This framework may support healthcare educators globally in integrating emerging AI technologies into their teaching, regardless of disciplinary focus.

Background: The use of Artificial Intelligence (AI) to analyze healthcare data has become common in behavioral health sciences. However, the lack of training opportunities for mental health professionals limits clinicians' ability to adopt AI in clinical settings. AI education is essential for trainees, equipping them with the literacy needed to implement AI tools in practice, collaborate effectively with data scientists, and develop as interdisciplinary researchers with computing skills. Objective: As part of the Penn Innovation in Suicide Prevention Implementation Research (INSPIRE) Center, we developed, implemented, and evaluated a virtual workshop to educate psychiatry and psychology trainees on using AI for suicide prevention research. Methods: The workshop introduced trainees to natural language processing (NLP) concepts and Python coding skills using jupyter notebooks within a secure Microsoft Azure Databricks cloud computing and analytics environment. We designed a three-hour workshop covered four key NLP topics: data characterization, data standardization, concept extraction, and statistical analysis. To demonstrate real-world applications, we processed chief complaints from electronic health record to compare the prevalence of suicide-related encounters across populations by race/ethnicity and age. Training materials were developed based on standard NLP techniques and domain-specific tasks, such as preprocessing psychiatry-related acronyms. Two researchers drafted and demonstrated the code, incorporating feedback from the INSPIRE Methods Core to refine the materials. To evaluate the workshop’s effectiveness, we used the Kirkpatrick program evaluation model, focusing on participants' reactions (Level 1) and learning outcomes (Level 2). Confidence changes in knowledge and skills before and after the workshop were assessed using paired t-tests, and open-ended questions were included to gather feedback for future improvements. Results: Ten attendees participated in the workshop virtually, including residents, postdocs, and graduate students from the psychiatry and psychology departments. Only two participants had experience with python or NLP prior to this workshop. They found the workshop helpful (mean = 3.17 on a scale of 1-4, SD = 0.41). Their overall confidence in NLP knowledge significantly increased (p < 0.002), from 1.35 (SD = 0.47) to 2.79 (SD = 0.46). Confidence in coding abilities also improved significantly (p = 0.013), increasing from 1.33 (SD = 0.60) to 2.25 (SD = 0.42). Open-ended feedback suggested incorporating theme analysis and exploring additional datasets for future workshops. Conclusions: This study illustrates the effectiveness of a tailored data science workshop for trainees in psychiatry and psychology, focusing on applying NLP techniques and suicide prevention research. The workshop significantly enhanced participants' confidence in conducting data science research. Future workshops will cover additional topics of interest, such as working with large language models, thematic analysis, diverse datasets, and multifaceted outcomes. This includes examining how participants' learning impacts their practice and research, as well as assessing knowledge and skills beyond self-reported confidence through methods like case studies for deeper insights. Clinical Trial: Not applicable

Background: Objective Structured Clinical Examinations (OSCEs) are used as an evaluation method in medical education #mededu, but require significant pedagogical expertise and investment, especially in emerging fields like digital health. Large language models (LLMs), such as ChatGPT (OpenAI), have shown potential in automating educational content generation. However, OSCE generation using LLMs remains underexplored. Objective: This study aims to evaluate 3 GPT-4o configurations for generating OSCE stations in digital health: (1) standard GPT with a simple prompt and OSCE guidelines; (2) personalized GPT with a simple prompt, OSCE guidelines, and a reference book in digital health; and (3) simulated-agents GPT with a structured prompt simulating specialized OSCE agents and the digital health reference book. Methods: Overall, 24 OSCE stations were generated across 8 digital health topics with each GPT-4o configuration. Format compliance was evaluated by one expert, while educational content was assessed independently by 2 digital health experts, blind to GPT-4o configurations, using a comprehensive assessment grid. Statistical analyses were performed using Kruskal-Wallis tests. Results: Simulated-agents GPT performed best in format compliance and most content quality criteria, including accuracy (mean 4.47/5, SD 0.28; P=.01) and clarity (mean 4.46/5, SD 0.52; P=.004). It also had 88% (14/16) for usability without major revisions and first-place preference ranking, outperforming the other configurations. Personalized GPT showed the lowest format compliance, while standard GPT scored lowest for clarity and educational value. Conclusions: Structured prompting strategies, particularly agents’ simulation, enhance the reliability and usability of LLM-generated OSCE content. These results support the use of artificial intelligence in medical education #mededu, while confirming the need for expert validation.

Background: Foundational knowledge of anesthesia techniques is essential for medical students. Team-based learning (TBL) improves engagement. Web-based virtual environments (WBVEs) allow many learners to join the same session in real time while being guided by an instructor. Objective: This study aimed to compare a WBVE with face-to-face (F2F) delivery of the same TBL curriculum in terms of postclass knowledge and learner satisfaction. Methods: We conducted a randomized, controlled, assessor-blinded trial at a Thai medical school from August 2024 to January 2025. Eligible participants were fifth-year medical students from the Faculty of Medicine, Khon Kaen University, who attended the anesthesiology course at the department of anesthesiology. Students who had previously completed the anesthesiology course or were unable to comply with the study protocol were excluded. They were allocated to one of the groups using a computer-generated sequence, with concealment of allocation to WBVE (on the Spatial platform) or F2F sessions. Both groups received identical 10-section content in a standardized TBL sequence lasting 130 minutes. Only the delivery mode differed (Spatial WBVE vs classroom F2F). The primary outcome was the postclass multiple-choice questionnaire score. The secondary outcome was learner satisfaction. Individual knowledge was assessed before and after the session using a 15-item questionnaire containing multiple-choice questions via Google Forms. Satisfaction was measured immediately after class on a 5-point Likert scale. Outcome scoring and data analysis were blinded to group assignment. Participants and instructors were not blinded. Results: In total, 79 students were randomized in this study (F2F: n=38, 48%; WBVE: n=41, 52%). We excluded 2% (1/41) of the students in the WBVE group due to incomplete data. There were complete data for the analysis for 78 participants (F2F: n=38, 49%; WBVE: n=40, 51%). Preclass scores were similar between groups (F2F: mean 6.03, SD 2.05; WBVE: mean 6.20, SD 2.04). Postclass knowledge did not differ significantly (F2F: mean 11.24, SD 1.93; WBVE: mean 10.40, SD 2.62; mean difference 0.88, 95% CI –0.18 to 1.94; P=.12). Learner satisfaction favored F2F learning across multiple domains, including overall course satisfaction. Overall satisfaction favored F2F learning (mean difference 0.42, 95% CI 0.07-0.77; P=.01). Both groups ran as planned. No adverse events were reported. No technical failures occurred in the WBVE group. Conclusions: In this trial, WBVE-delivered TBL produced similar short-term knowledge gains to F2F delivery, but learner satisfaction was lower in the WBVE group. Unlike many previous studies, this trial compared WBVE and F2F delivery while keeping the TBL curriculum and prespecified outcomes identical across groups. These findings support WBVEs as a scalable option when physical space, learner volume, or constraints are present. However, lower satisfaction in the WBVE highlights the real-world need for improved facilitation, user experience design, and technical readiness before broader implementation. Clinical Trial: Thai Clinical Trials Registry TCTR20240708012; https://www.thaiclinicaltrials.org/show/TCTR20240708012

Background: The rapid advancement of artificial intelligence (AI) has had a substantial impact on medicine, necessitating the integration of AI education into medical school curricula. However, such integration remains limited. A key challenge is the discrepancy between medical students’ positive perceptions of AI and their actual competencies, with research in Japan identifying specific gaps in the students’ competencies in understanding regulations and discussing ethical issues. Objective: This study evaluates the effectiveness of an educational program designed to improve medical students’ competencies in understanding legal and ethical AI-related issues. It addresses the following research questions: (1) Does this educational program improve students’ knowledge of AI and its legal and ethical issues, and what is each program element’s contribution to this knowledge? (2) How does this educational program qualitatively change medical students’ thoughts on these issues from an abstract understanding to a concrete and structured thought process? Methods: This mixed methods study used a single-group pretest and posttest framework involving 118 fourth-year medical students. The 1-day intervention comprised a lecture and problem-based learning (PBL) session centered on a clinical case. A 24-item multiple-choice questionnaire (MCQ) was administered at 3 time points (pretest, midtest, and posttest), and descriptive essays were collected before and after the intervention. Data were analyzed using linear mixed-effects models, the Wilcoxon signed-rank test, and text mining, including comparative frequency analysis and cooccurrence network analysis with Jaccard coefficients. An optional survey on student perceptions based on the attention, relevance, confidence, and satisfaction model was conducted (n=76, 64.4%). Results: Objective knowledge scores increased significantly from the pretest (median 17, IQR 15-18) to posttest (median 19, IQR 17-21; β=1.42; P

Background: In the current era of artificial intelligence (AI), utilization of AI has increased in both clinical practice and medical education #mededu. Nevertheless, it is probable that perspectives on the prospects and risks of AI vary among individuals. Given the potential for attitudes toward AI to significantly influence its integration into medical practice and educational initiatives, it is essential to assess these attitudes using a validated tool. The recently developed 12-item Attitude towards Artificial Intelligence (ATTARI-12) scale has demonstrated good validity and reliability for the general populations, suggesting its potential for extensive utilization in future studies. However, to our knowledge, there is currently no validated Japanese version of the scale. The lack of a Japanese version hinders research and educational efforts aimed at understanding and improving AI integration into the Japanese healthcare and medical education #mededu system. Objective: We aimed to develop the Japanese version of the ATTARI-12 (J-ATTARI-12) scale and investigate whether it is applicable to medical trainees. Methods: We first translated the original English scale into Japanese. To examine its psychometric properties, we then conducted a validation survey by distributing the translated version as an online questionnaire to medical students and residents across Japan from June to July 2025. We assessed structural validity through factor analysis, and convergent validity by computing the Pearson correlation coefficient between the J-ATTARI-12 scale scores and the attitudes towards robots scores. Internal consistency reliability was assessed by Cronbach’s alpha values. Results: We included 326 participants in our analysis. We employed a split-half validation approach, with exploratory factor analysis (EFA) on the first half and confirmatory factor analysis (CFA) on the second. EFA suggested a two-factor solution (Factor 1, AI anxiety and aversion; Factor 2, AI optimism and acceptance). CFA revealed that the model fitness indices of the two-factor structure suggested by the EFA was good (comparative fit index: 0.914 (> 0.900); root mean square error of approximation: 0.075 (< 0.080); standardized root mean square residual 0.056 (< 0.080)) and superior to those of the one-factor structure. The value of the Pearson correlation coefficient between the J-ATTARI-12 scale scores and attitudes towards robots scores was 0.52, which indicated good convergent validity. Cronbach’s alpha value for all 12 items was 0.84, which indicated a high level of internal consistency reliability. Conclusions: We developed and validated the J-ATTARI-12 scale. The developed instrument had good structural validity, convergent validity, and internal consistency reliability for medical trainees. The J-ATTARI-12 scale is expected to stimulate future studies and educational initiatives that can effectively assess and enhance the integration of AI into clinical practice and medical education #mededu systems.

Background: As medical and allied health curricula adapt to increasing time constraints, ethical considerations, and resource limitations, digital innovations are becoming vital supplements to donor-based anatomy instruction. While prior studies have examined the effectiveness of prosection versus dissection and the role of digital tools in anatomy learning, few resources align interactive digital modules directly with hands-on prosection experiences. Objective: This project addresses that gap by introducing an integrated, curriculum-aligned platform for self-guided cadaveric learning. Methods: We created Anatomy Interactives, a web-based laboratory manual structured to complement prosection laboratories for MD, DPT, and PA students. Modules were developed using iSpring Suite (iSpring Solutions Incorporated) and included interactive labeled images, donor photographs, and quiz-style self-assessments. Learners engaged with modules before, during, or after laboratory sessions. PA/DPT and MD students completed postcourse surveys evaluating module use and perceived impact. MD student examination scores from a 2023 cohort (no module access) were compared to a 2024 cohort (with access) to evaluate effectiveness. Results: A total of 147 students completed the survey (31 PA/DPT and 116 MD). The majority reported using modules for 1-2 hours per week and found them helpful for both written and laboratory examinations. MD students in the 2024 cohort performed better on all 3 examinations compared to the 2023 cohort, with 2 examination median differences reaching statistical significance (Mann-Whitney U, P

Background: In recent years, generative artificial intelligence and large language models (LLMs) have rapidly advanced, offering significant potential to transform medical education #mededu. Several studies have evaluated the performance of chatbots on multiple-choice medical exams. Objective: The study aims to assess the performance of two LLMs – GPT-4o and OpenAI o1 – on the Médico Interno Residente (MIR) 2024 exam, the Spanish national medical test that determines eligibility for competitive medical specialist training positions. Methods: A total of 176 questions from the MIR 2024 exam were analysed. Each question was presented individually to the chatbots to ensure independence and prevent memory retention bias. No additional prompts were introduced to minimize potential bias. For each LLM, response consistency under verification prompting was assessed by systematically asking, "Are you sure?" after each response. Accuracy was defined as the percentage of correct responses compared to the official answers provided by the Spanish Ministry of Health. It was assessed for GPT-4o, OpenAI o1 and, as a benchmark, for a consensus of medical specialists and for the average MIR candidate. Sub-analyses included performance across different medical subjects, question difficulty (quintiles based on the percentage of examinees correctly answering each question), and question types (clinical cases versus theoretical questions; positive versus negative questions). Results: Overall accuracy was 158/176 (89.8%) for GPT-4o - and 160/176 (90.0%) after verification prompting, 163/176 (92.6%) for OpenAI o1 - and 164/176 (93.2%) after verification prompting, 166/176 (94.3%) for the consensus of medical specialists, and 100/176 (56.6%) for the average MIR candidate. Both LLMs and the consensus of medical specialists outperformed the average MIR candidate across all 20 medical subjects analyzed, with ≥80% LLMs’ accuracy in most domains. A performance gradient was observed: LLMs’ accuracy gradually declined as question difficulty increased. Slightly higher accuracy was observed for clinical cases compared to theoretical questions, as well as for positive questions compared to negative ones. Both models demonstrated high response consistency, with near-perfect agreement between initial responses and those after the verification prompting. Conclusions: These findings highlight the excellent performance of GPT-4o and OpenAI o1 on the MIR 2024 exam, demonstrating consistent accuracy across medical subjects and question types. The integration of LLMs into medical education #mededu presents promising opportunities and is likely to reshape how students prepare for licensing exams and change our understanding of medical education #mededu. Further research should explore how wording, language, prompting techniques and image-based questions can influence LLMs’ accuracy, as well as evaluate the performance of emerging artificial intelligence (AI) models in similar assessments.

Background: Ultrasound-guided regional anesthesia (UGRA) remains underused in low- and middle-income countries due to barriers to training and equipment. Recent advances in portable ultrasound devices and international partnerships have expanded access to UGRA, enhancing patient safety and quality of care. Objective: This study describes the development and outcomes of a hybrid UGRA training program for anesthesiologists at the Hospital Nacional de Coatepeque (HNC) in Guatemala. Methods: An educational pilot program for UGRA was developed based on local needs and feedback, comprising 4 weeks of online modules, an in-person educational conference, and 1 month of supervised clinical practice. Evaluation followed the Kirkpatrick framework using preprogram and postprogram surveys adapted from the Global Regional Anesthesia Curricular Engagement model. Outcomes included participants’ satisfaction, change in knowledge and skill, and procedural performance. Knowledge and skill assessments were compared before and after the training, and clinical data were recorded for 10 months. Nonparametric tests were used to assess changes and associations with performance outcomes. Results: All 7 anesthesiologists at HNC completed the training program. Knowledge test scores improved by a median percentage increase of 20.8% (IQR 13.5%-28.1%; r=0.899; P=.02), and procedural skill rating scores increased by a median percentage of 147.1% (IQR 96.9%-197.3%; r=0.904; P=.03) at 1 month and 131.4% (IQR 90.5%-172.3%; r=0.909; P=.04) at 4 months after the program. Participants self-reported high satisfaction and substantial clinical improvement and motivation. A total of 54 peripheral nerve blocks were performed under direct supervision in the first month, with 187 blocks recorded over 10 months. The supraclavicular brachial plexus block was the most frequently used (66/187, 35.3%) and replaced the standard general anesthetic for upper extremity surgery in 70 patients. The procedure success rate was 96.3% (180/187), and there were no observed patient complications. Conclusions: This hybrid curriculum enabled the successful implementation of UGRA at a public hospital in Guatemala, safely expanding clinical capabilities and reducing reliance on general anesthesia for upper extremity surgery. This practical training model provides a framework for implementing UGRA in similar resource-limited hospitals.

Background: Artificial intelligence (AI) literacy is increasingly essential for medical students. However, without systematic characterization of the relevant components, designing targeted medical education #mededu interventions may be challenging. Objective: Systematically describe (1) the levels of and (2) factors associated with multidimensional AI literacy among Chinese medical students. Methods: A cross-sectional, descriptive analysis was conducted using data from a nationwide survey of Chinese medical students (n = 80,335) across 109 medical schools in 2024. AI literacy was assessed with a multidimensional instrument comprising three domains: knowledge, evaluating students’ self-reported proficiency in core areas of medical AI applications; attitude, reflecting their self-perceived views on using AI for teaching and learning; and behavior, capturing the self-perceived usage frequency and application patterns. Multivariate linear regression was applied to examine the associations between individual factors (i.e., demographic characteristics, family background, and enrollment motivation) and environmental factors (i.e., educational phase, type of education program, and tier of education program) and AI literacy. Results: Respondents showed moderate to high levels of AI knowledge (mean, 76.0 [SD, 26.9]), followed by moderate AI attitude scores (mean, 71.6 [SD, 24.4]). In contrast, AI behavior scores were much lower (mean, 32.5 [SD, 28.5]), indicating little usage of AI tools. Of the individual factors, male students reported higher levels of AI attitude and behavior; both intrinsic and extrinsic motivation were positively associated with all three dimensions; advantaged family background was positively related to AI attitude and behavior, but not knowledge. Among the environmental factors, attending the prestigious Double First-Class universities was positively associated with higher AI usage. Enrollment in long-track medical education #mededu programs was associated with higher AI attitude and behavior, while being in the clinical phase was negatively associated with both AI knowledge and behavior. Environmental factors moderated the associations between individual characteristics and AI literacy, potentially attenuating disparities. Conclusions: Medical students reported moderate to high AI knowledge, moderate AI favorability, and low AI use. Individual characteristics and environmental factors were significantly associated with AI literacy, and environmental factors moderated the associations. The moderate AI literacy overall highlights the need for AI-related medical education #mededu, ideally with practical use and nuanced by socioeconomic factors. Clinical Trial: Not applicable

Background: Artificial Intelligence (AI), particularly large language models (LLMs) like ChatGPT, is rapidly influencing medical education #mededu. Its effectiveness for students with varying levels of prior knowledge remains underexplored Objective: This study aimed to evaluate the performance of medical students with and without formal pharmacology knowledge when using AI-LLM GPTs, internet search engines, e-books, or self-knowledge to solve multiple-choice questions (MCQs). Methods: A cross-sectional study was conducted with 100 medical students, divided into a "naïve" group (n=50, no pharmacology training) and a "learned" group (n=50, completed pharmacology training). Each participant answered four sets of 20 MCQs using self-knowledge, e-books, Google, or ChatGPT-4o. Scores were compared using ANCOVA, with self-knowledge scores as a covariate. Results: Learned students significantly outperformed naïve students across all methods (P < 0.001), with the largest effect size in the AI-LLM GPT set (partial η² = 0.328). For both groups, the performance hierarchy was AI-LLM GPT > Internet search engine > Self-knowledge ≈ E-books. Crucially, naïve students using AI (mean=13.24, SD=3.31) scored higher than learned students using Google (mean=12.14, SD=2.01; P=0.0107) or e-books (mean=10.22, SD=3.12; P

Background: Podcasts are increasingly used in health professions education, yet most formats are asynchro-nous and non-interactive. Didactically grounded, synchronous implementations in dental cur-ricula are scarce. Objective: To design, implement, and evaluate a synchronous, case-based Live Podcast (LP) as a didactic teaching format in dentomaxillofacial radiology (DMFR). Methods: In a controlled cohort study with two third-year cohorts (n = 41), the intervention group (IG; 21/41) received weekly case-based Live Podcast sessions in addition to standard teaching, while the control group (CG; 20/41) received standard teaching only. Acceptability was eval-uated six months post-course using the 27-item Student Evaluation Questionnaire (SEQ) and open-text responses. Knowledge was assessed immediately after the course with a 21-item Radiology Knowledge Test (RKT), and after six months with a 15-item Interdisciplinary Clin-ical Application Test (ICAT). Results: The primary outcome was student-reported acceptability of the LP format, it was rated highly by students in the SEQ (mean out of 10: structure 9.76, interactivity 9.62, interdisciplinary relevance 9.55). Qualitative feedback was assessed highlighting motivation, authenticity, and discussion quality. In the RKT, no group differences were observed (IG 51.2% vs. CG 48.8%; P=.37). In the ICAT, the IG outperformed the CG in restorative dentistry (median 5 vs. 4; P=.02, r=0.38) and in item-level analysis (71% vs. 40%; P=.04, φ=0.64). Conclusions: The LP format represents a feasible, scalable, and low-threshold approach to fostering clinical reasoning in dental curricula, particularly at the transition to clinical training. While radiology-specific theoretical competences did not differ between groups, students consistently rated the LP as more engaging and motivating compared to standard lectures

Background: Problem-Based Learning (PBL) has gained widespread acceptance in medical education #mededu. Given WeChat has emerged as a popular social networking platform in China, we have opted to utilize it for conducting online PBL in ophthalmology, aiming to diminish the constraints of conventional teaching approaches. Objective: This study aims to assess the effectiveness of problem-based learning (PBL) combined with WeChat in Chinese undergraduate medical students, compared to traditional teaching methods. Methods: The study involved a total of 108 undergraduate students who successfully passed the National Entrance Examination. Students were placed into six groups (18 students for each group) using a random number table, and the new teaching methods were tested outside their regular class time. Three groups were randomly selected to receive PBL using WeChat as the platform, while the remaining three groups received conventional teaching. Results: Total scores were not significantly different between WeChat-based PBL teaching group and traditional teaching group; the memory scores, one of the component of total scores, were significantly higher in traditional teaching group compared to WeChat-based PBL group. Conclusions: Compared to conventional teaching, WeChat-based PBL model is more easily accepted by students, and has transformed the traditional "cramming education" model by facilitating a more active learning experience for students.

Background: Artificial intelligence (AI) is rapidly changing both clinical psychiatry and the education of medical professionals. However, little is currently known about how AI is being discussed in the education and training of psychiatry for medical students and doctors around the world. Objective: This paper aims to provide a snapshot of the available data on this subject as of 2024. A deliberately broad definition of AI was adopted to capture the widest range of relevant literature and applications, including machine learning, natural language processing, and generative AI tools. Methods: A scoping review was conducted using both peer-reviewed publications from PubMed, Embase, PsycINFO and Scopus databases, and grey literature resources. The criterion for inclusion was a description of how AI could be applied to education or training in psychiatry. Results: A total of 26 records published between 2016 and 2024 were included. The key themes identified were (i) the imperative for an AI curriculum for students or doctors training in psychiatry; (ii) uses of AI to develop educational resources; (iii) uses of AI to develop clinical skills; (iv) uses of AI for assessments; (v) academic integrity or ethical considerations surrounding the use of AI; and (vi) tensions relating to competing priorities and directions. Conclusions: Although a nascent field, it is clear that AI will increasingly impact on assessment, clinical skills training, and the development of teaching resources in psychiatry. Training curricula will need to reflect the new knowledge and skills required for future clinical practice. Educators will need to be mindful of academic integrity risks and to emphasise development of critical thinking skills. Attitudes of psychiatrists toward the rise of AI in training remain underexplored.

Background: While artificial intelligence (AI)–generated feedback offers significant potential to overcome constraints on faculty time and resources associated with providing personalized feedback, its perceived usefulness can be undermined by algorithm aversion. In-context learning, particularly the few-shot approach, has emerged as a promising paradigm for enhancing AI performance. However, there is limited research investigating its usefulness, especially in health profession education. Objective: This study aimed to compare the quality of AI-generated formative feedback from 2 settings, feedback generated in a zero-shot setting (hereafter, “zero-shot feedback”) and feedback generated in a few-shot setting (hereafter, “few-shot feedback”), using a mixed methods approach in Japanese physical therapy education. Additionally, we examined the effect of algorithm aversion on these 2 feedback types. Methods: A mixed methods study was conducted with 35 fourth-year physical therapy students (mean age 21.4, SD 0.7 years). Zero-shot feedback was created using Gemini 2.5 Pro with default settings, whereas few-shot feedback was generated by providing the same model with 9 teacher-created examples. The participants compared the quality of both feedback types using 3 methods: a direct preference question, the Feedback Perceptions Questionnaire (FPQ), and focus group interviews. Quantitative comparisons of FPQ scores were performed using the Wilcoxon signed rank test. To investigate algorithm aversion, the study examined how student perceptions changed before and after disclosure of the feedback’s identity. Results: Most students (26/35, 74%) preferred few-shot feedback over zero-shot feedback in terms of overall usefulness, although no significant difference was found between the 2 feedback types for the total FPQ score (P=.22). On the specific FPQ scales, few-shot feedback scored significantly higher than zero-shot feedback on fairness across all 3 items: “satisfied” (P=.02; r=0.407), “fair” (P=.04; r=0.341), and “justified” (P=.02; r=0.392). It also scored significantly higher on 1 item of the usefulness scale (“useful”; P=.02; r=0.401) and 1 item of the willingness scale (“invest a lot of effort”; P=.02; r=0.394). In contrast, zero-shot feedback scored significantly higher on the affect scale across 2 items: “successful” (P=.03; r=0.365) and “angry” (P=.008; r=0.443). Regarding algorithm aversion, evaluations for zero-shot feedback became more negative for 83% (15/18) of the items after identity disclosure, whereas positive perceptions of few-shot feedback were maintained or increased. Qualitative analysis revealed that students valued zero-shot feedback for its encouraging tone, whereas few-shot feedback was appreciated for its contextual understanding and concrete guidance for improvement. Conclusions: Japanese physical therapy students perceived few-shot feedback more favorably than zero-shot feedback on case reports. This few-shot AI model shows potential to resist algorithm aversion and serves as an effective educational tool to support autonomous writing, facilitate reflection on clinical reasoning, and cultivate advanced thinking skills.

Background: The integration of artificial intelligence (AI) and machine learning (ML) into biomedical research requires a workforce fluent in both computational methods and clinical applications. Structured, interdisciplinary training opportunities remain limited, creating a gap between data scientists and clinicians. The National Institutes of Health’s Bridge2AI initiative launched the Artificial Intelligence–Ready and Exploratory Atlas for Diabetes Insights (AI-READI) Data Generation Project to address this gap. AI-READI is creating a multimodal, FAIR (Findable, Accessible, Interoperable, and Reusable) dataset—including ophthalmic imaging, physiologic measurements, wearable sensor data, and survey responses—from approximately 4,000 participants with or at risk for type 2 diabetes. In parallel, AI-READI established a yearlong mentored research program that begins with a two-week immersive summer bootcamp to provide foundational AI/ML skills grounded in domain-relevant biomedical data. Objective: To describe the design, iterative refinement, and outcomes of the AI-READI Bootcamp, and to share lessons for creating future multidisciplinary AI/ML training programs in biomedical research. Methods: Held annually at UC San Diego, the bootcamp combines 80 hours of lectures, coding sessions, and small-group mentorship. Year 1 introduced Python programming, classical ML techniques (e.g., logistic regression, convolutional neural networks), and data science methods such as principal component analysis and clustering, using public datasets. In Year 2, the curriculum was refined based on structured participant feedback—reducing cohort size to increase individualized mentorship, integrating the AI-READI dataset (including retinal images and structured clinical variables), and adding modules on large language models and FAIR data principles. Participant characteristics and satisfaction were assessed through standardized pre- and post-bootcamp surveys, and qualitative feedback was analyzed thematically by independent coders. Results: Seventeen participants attended Year 1 and seven attended Year 2, with an instructor-to-student ratio of approximately 1:2 in the latter. Across both years, post-bootcamp evaluations indicated high satisfaction, with Year 2 participants reporting improved experiences due to smaller cohorts, earlier integration of the AI-READI dataset, and greater emphasis on applied learning. In Year 2, mean scores for instructor effectiveness, staff support, and overall enjoyment were perfect (5.00/5.00). Qualitative feedback emphasized the value of working with domain-relevant, multimodal datasets; the benefits of peer collaboration; and the applicability of skills to structured research projects during the subsequent internship year. Conclusions: The AI-READI Bootcamp illustrates how feedback-driven, multidisciplinary training embedded within a longitudinal mentored research program can bridge technical and clinical expertise in biomedical AI. Core elements—diverse trainee cohorts, applied learning with biomedical datasets, and sustained mentorship—offer a replicable model for preparing health professionals for the evolving AI/ML landscape. Future iterations will incorporate additional pre-bootcamp onboarding modules, objective skill assessments, and long-term tracking of research engagement and productivity.

Date Submitted: Dec 23, 2025. Open Peer Review Period: Dec 23, 2025 - Feb 17, 2026.

Date Submitted: Dec 23, 2025. Open Peer Review Period: Dec 23, 2025 - Feb 17, 2026.

Background: Technological innovation is reshaping the landscape of medical education #mededu, bringing revolutionary changes to traditional teaching methods. In this context, the upgrade of the teaching model for microscopy, as one of the core skills in medical education #mededu, is particularly important. Proficiency in microscope operation not only affects medical students’ pathology diagnosis abilities but also directly impacts the precision of surgical procedures and laboratory analysis skills. However, current microscopy pedagogy faces dual challenges: on one hand, traditional teaching lacks real-time image sharing capabilities, severely limiting the effectiveness of immediate instructor guidance; on the other hand, students find it difficult to independently identify technical flaws in their operations, leading to inefficient skill acquisition. Although whole-slide imaging-based microscopy system technology has partially addressed the issue of image visualization, it cannot replicate the tactile feedback and physical interaction experience of the real world. The breakthrough development of 5G communication technology—with its ultrahigh transmission speed and ultralow latency—provides an innovative solution to this teaching challenge. Leveraging this technological advantage, Tongji University’s biology laboratory has pioneered the deployment of a 5G local area network (LAN)–supported digital interactive microscopy system, creating a new model for microscopy education. Objective: This study aims to investigate the efficacy of an innovative 5G LAN-powered interactive digital microscopy system in enhancing microscopy training efficiency, evaluated through medical students’ academic performance and learning experience. Methods: Using a quasi-experimental design, we quantify system effectiveness via academic performance metrics and learning experience dimensions. A total of 39 students enrolled in the biology course were randomly assigned to 2 groups: one using traditional optical microscopes (control) and the other using the digital microscopy interactive system (DMIS). Their academic performance was evaluated through a knowledge test and 3 laboratory reports. A 5-point Likert-scale questionnaire was used to gather feedback on students’ learning experiences. In addition, the DMIS group was required to evaluate the specific functions of the system. Results: In the knowledge test, no statistical difference was found between the 2 groups; however, the DMIS group scored significantly higher in Lecture 2 (P

Background: The Japanese National Medical Licensing Examination (NMLE) is mandatory for all medical graduates seeking to become licensed physicians in Japan. Given the cultural emphasis on summative assessment, the NMLE has had a significant impact on Japanese medical education #mededu. Although the NMLE Content Guidelines have been revised approximately every five years over the last 2 decades, objective literature analyzing how the examination itself has evolved is absent. Objective: To provide a holistic view of the trends of the actual examination over time, this study used a combined rule-based and data-driven approach. We primarily focused on classifying the items according to the perspectives outlined in the NMLE Content Guidelines, complementing this approach with a natural language processing technique called topic modeling to identify latent topics. Methods: We collected publicly available NMLE data for 2001-2024. Six examination iterations (2880 items) were manually classified from 3 perspectives (level, content, and taxonomy) based on pre-established rules derived from the guidelines. Temporal trends within each classification were evaluated using the Cochran-Armitage test. Additionally, we conducted topic modeling for all 24 examination iterations (11,540 items) using the bidirectional encoder representations from transformers–based topic modeling framework. Temporal trends were traced using linear regression models of topic frequencies to identify topics growing in prominence. Results: In the level classification, the proportion of items addressing common or emergent diseases increased from 60% (115/193) to 76% (111/147; P