Latest Submissions Open for Peer Review

Titles/Abstracts of Articles Currently Open for Review:

• Patterns of Artificial Intelligence Use for Studying Among Medical Students: A Longitudinal Digital Diary Study

  Date Submitted: Apr 1, 2026

  Open Peer Review Period: Apr 2, 2026 - May 28, 2026

  • Peer Review Me

  Background: Artificial intelligence (AI) is increasingly integrated into medical education, offering new ways for students to acquire knowledge and support clinical reasoning. However, the extent, patterns, and implications of AI use among medical students remain incompletely understood. Objective: This study aimed to quantify real-time artificial intelligence use among medical students, including the proportion of study time devoted to AI, and to examine how use varies by training stage and engagement style (active vs passive). Methods: This longitudinal study recruited medical students from two osteopathic medical schools to complete a baseline survey and seven digital diary entries over a 21-day period. The diary method was designed to capture real-time AI use and reduce recall bias. Data were analyzed using Stata 19. Multiple linear regression models examined associations between AI use (total minutes and percentage of study time) and key variables, including year in training and type of use (active vs passive), adjusting for age, gender, and campus. Results: A total of 71 students completed the baseline survey (mean age 26.6 years, SD 2.8; 54.9% male; 77.4% pre-clinical). On average, students reported using AI tools during 19.4% of their total study time. Clinical-phase students (MS3–MS4) used AI significantly more than pre-clinical students (MS1–MS2), with an adjusted increase of 19.0 percentage points (p=0.003). Students classified as active users spent significantly more total time using AI than passive users (p=0.002). Across groups, AI use was primarily passive, including simplifying complex concepts, answering practice questions, and generating summaries. Clinical-phase students were more likely to use AI for practice questions. Conclusions: Medical students are incorporating AI into a substantial proportion of their study time, with greater use among clinical trainees and those engaging actively with these tools. Despite this, most use remains passive. Given mixed evidence regarding the impact of AI on deep learning and potential risks related to uncritical acceptance of AI-generated content, these findings highlight the need for further research on learning outcomes. Medical schools may benefit from providing guidance on responsible AI use, including critical evaluation, verification of outputs, and integration into evidence-based study strategies. The digital diary methodology offers a novel and practical approach for capturing real-time AI use and may inform future educational research and intervention design.

• How quickly can the medical curriculum adapt to rapidly changing knowledge: a qualitative study using the example of digitalization in medicine

  Date Submitted: Mar 26, 2026

  Open Peer Review Period: Mar 30, 2026 - May 25, 2026

  • Peer Review Me

  Background: Digitalization is transforming the way we provide and experience medicine and healthcare. Experts have suggested various topics for medical curricula to keep pace with rapidly evolving knowledge; however, adapting these curricula remains a lengthy process that often lacks an interdisciplinary approach. Objective: This study examines the perspective of curriculum governing bodies and the boards responsible for curriculum operations at two medical universities towards the need for necessary curriculum changes to account for digitalization in medicine and the difficulties in adapting the curriculum to ever-growing knowledge. We identify and suggest ways for a more agile curriculum development. Methods: This study consists of an qualitative analysis of governing policy frameworks and a qualitative study involving 14 video interviews. The interviews were performed with members of university curriculum governing bodies and the boards responsible for curriculum operations. Results: Participants agreed that digitalization will reshape the medical profession by reducing physical contact, enhancing data-driven communication, and streamlining administrative processes. They highlighted the need for graduates to acquire digital literacy, critical evaluation skills, and a basic understanding of data and statistics. Yet, despite being designed as integrated program, participants noted curricula have become fragmented over time due to missing coordination between curriculum modules. Furthermore, current processes lead to a siloed perspective, where limited coordination between modules makes it difficult to implement new knowledge holistically. This lack of inter-module alignment emerged as a key barrier to coherent curricular change. Learning objectives were identified as a promising but underutilized tool for monitoring content, aligning modules, and ensuring that emerging topics like digitalization are integrated consistently. Conclusions: Participants agreed that current processes for monitoring and updating curricula are not efficiently designed and tend to be too static and focus on the advancement of subject-specific medical knowledge. To prepare current and future students for a rapidly changing world, curriculum processes should evolve from static, fragmented structures to more agile, integrated systems. By mapping the survey results to the curriculum development frameworks of Kern and Harden, we find that the challenge lies not so much in adding new content, but rather in designing curriculum processes that achieve a holistic overview. Strengthening the use of learning objectives as a dynamic monitoring and alignment tool offers a concrete opportunity to integrate rapidly changing knowledge holistically.

• Lecturer-in-the-Loop Clinical Dialectic (LLCD): A Framework for AI-Mediated Socratic Simulation in Resource-Limited Settings

  Date Submitted: Mar 25, 2026

  Open Peer Review Period: Mar 26, 2026 - May 21, 2026

  • This manuscript needs more reviewers Peer Review Me

  Traditional clinical simulation requires substantial infrastructure investment, limiting accessibility in resource-constrained settings. AI technologies hold promise for scalable simulation yet concerns about clinical accuracy and faculty displacement remain. We describe the “Lecturer-in-the-Loop” Clinical Dialectic (LLCD), a framework integrating AI-mediated Socratic simulation with faculty oversight and share our initial experience with final-year medical students at an academic hospital in Jamaica. We facilitated a teaching session with seven final-year medical students using Claude Opus (Anthropic). Students engaged with two sequential AI-generated pediatric cases: acute asthma exacerbation, then bronchiolitis. These clinical scenarios evolved dynamically based on student decisions. Through dialogue with the system, students asked questions, consolidated pathophysiology, proposed management plans, and requested clarification and elaboration on recommendations. Notably, students independently applied pharmacological reasoning from the asthma case to determine that bronchodilators were inappropriate for bronchiolitis, an unprompted transfer of mechanistic understanding across cases. Faculty provided continuous oversight: prompting students to articulate their clinical reasoning before committing to answers, reinforcing key learning points, and validating AI-generated content in real-time. When the AI generated equivocal or clinically inaccurate content, faculty insight transformed these moments into teaching opportunities about critical appraisal. The session ran approximately 3 hours with sustained student engagement. LLCD may represent a reproducible, low-cost approach to clinical simulation-based education that preserves the central role of faculty while leveraging AI’s dialogic capabilities. By positioning AI as a dialogic tool requiring expert validation rather than an autonomous teacher, the framework addresses safety concerns while enabling scalable simulation in resource-limited settings where high-fidelity simulation infrastructure remains inaccessible.

• Assessment of Verbal Communication in Dementia Care: A Comparative Study of Large Language Models and Human Annotations

  Date Submitted: Mar 25, 2026

  Open Peer Review Period: Mar 26, 2026 - May 21, 2026

  • This manuscript needs more reviewers Peer Review Me

  Background: Effective verbal communication is a core component of nursing care, particularly in dementia care such as Humanitude. However, manual evaluation of communication quality is time-consuming, subjective, and difficult to scale in training settings. Large language models (LLMs) may enable automated and scalable analysis of verbal communication in caregiving. Objective: This study evaluated whether LLMs can reliably classify verbal communication in nursing care training sessions and detect differences in communication patterns across caregiver expertise levels. Methods: Care sessions involving simulated patients were conducted with 18 participants, including Humanitude instructors, intermediate practitioners, and novice nurses. Audio recordings were transcribed, segmented into utterances, and classified into 6 communication categories: positive/affectionate expression, request/suggestion, gratitude, explanation, question/confirmation, and none. Four human annotators independently labeled the utterances, and the same transcripts were analyzed using GPT, Claude, and Gemini. Agreement was evaluated using pairwise agreement rates and Cohen’s kappa coefficients. Model performance was further assessed against consensus labels derived from multiple annotators, and non-inferiority/equivalence was tested using two one-sided tests (TOST). Results: Inter-annotator agreement among the human annotators was moderate, with pairwise agreement rates ranging from 64.44% to 74.21% and Cohen’s kappa values ranging from 0.554 to 0.664. Among the evaluated LLMs, Claude showed the highest agreement with human annotations, followed by Gemini and GPT. Against consensus labels, Claude achieved the highest accuracy (0.836 for ≥2-annotator consensus; 0.902 for ≥3-annotator consensus), followed by Gemini (0.779; 0.837) and GPT (0.672; 0.732). TOST analysis showed that Gemini achieved statistical equivalence with human annotation (p=0.040), while Claude demonstrated non-inferiority and exceeded the human baseline (p=0.001). Across caregiver groups, instructors showed a higher proportion of positive/affectionate expressions, whereas novice caregivers showed a higher proportion of task-oriented and uncategorized utterances. Overall, LLM-based classification reproduced the general communication patterns observed in human annotations. Conclusions: LLM-based classification demonstrated reliability comparable to human annotation for caregiving communication analysis. Claude showed the strongest overall performance, and Gemini achieved statistical equivalence with human annotation. These findings suggest that LLM-based analysis may provide a scalable and objective approach to assessing communication behaviors in Humanitude training and support communication assessment in nursing and medical education. Clinical Trial: Gunma University Hospital (HS2024-044)

• Effectiveness and Students’ Perception of Tutor-Guided AI Navigation in Undergraduate Medical Education in Sri Lanka: A Quasi-Experimental Study

  Date Submitted: Mar 22, 2026

  Open Peer Review Period: Mar 22, 2026 - May 17, 2026

  • This manuscript needs more reviewers Peer Review Me

  Background: The integration of Artificial Intelligence (AI), particularly Large Language Models (LLMs) such as ChatGPT, into undergraduate medical education offers new opportunities for personalized learning. However, concerns remain regarding content accuracy, possible over-reliance and limited critical engagement when AI tools are used without guidance. Tutor-guided AI navigation has been proposed as a structured approach combining AI accessibility with educator oversight. Objective: This study aimed to evaluate the effectiveness and perception of tutor-guided AI navigation among medical undergraduates. Methods: The study was conducted among 87 final-year medical students at the Faculty of Medicine, University of Kelaniya, Sri Lanka. A quasi-experimental pre-test/post-test design compared knowledge gains following tutor-guided versus self-directed ChatGPT-assisted learning. A cross-sectional survey assessing three domains-Perception of Learning Experience, AI Usability and Confidence, Satisfaction and Overall Impression- was administered immediately after the session. Responses were measured using a five-point Likert scale. Internal consistency was evaluated with Cronbach’s alpha and descriptive statistics were analysed using SPSS. Results: Baseline pre-test scores were comparable between groups (W = 670.5, p = 0.153). Both groups improved significantly from pre- to post-test (p < 0.001), with median scores increasing from 4 to 7. Knowledge gain between groups was not statistically significant (Welch’s t = 1.07, p = 0.144), although the effect size (Hedges’ g = 0.36) suggested a small-to-moderate advantage for tutor guidance. In the perception survey, reliability was high (Cronbach’s alpha > 0.86). Students rated ease of use (4.19 ± 1.07), tutor-guided AI experience (4.03 ± 1.13), satisfaction (3.91 ± 0.94) and willingness to recommend (4.01 ± 0.97) favourably, while trust in AI accuracy was moderate (3.27 ± 0.91). Conclusions: Tutor-guided AI-assisted learning was well accepted by students and provided a positive, structured learning experience. Although it did not produce a statistically significant advantage in immediate knowledge gain over self-directed AI use, it was associated with favourable perceptions of usability, satisfaction and recommendation. These findings support the structured integration of tutor-guided AI into undergraduate medical education while maintaining educator oversight and critical appraisal of AI-generated content.

• Using Generative Artificial Intelligence to Aid in Surgery Resident Selection: A Retrospective Comparative Study

  Date Submitted: Mar 16, 2026

  Open Peer Review Period: Mar 18, 2026 - May 13, 2026

  • This manuscript needs more reviewers Peer Review Me

  Background: Surgery resident selection is a resource-intensive process. The advent of generative artificial intelligence (GAI) offers a new possibility to aid in resident selection, increasing the efficiency of file review without the burden of creating a customized machine-learning algorithm. Objective: Our study aimed to compare file review of general surgery applicants by GAI to file review by our program’s residency selection committee (RSC). Methods: GPT-4o, an open access GAI software, was used to score deidentified 2023-2024 Canadian Resident Matching Service (CaRMS) application files to our program based on our RSC’s file review scoresheet. GAI scores were compared to RSC-assigned scores for each application element including CVs, personal letters, and reference letters. Rank lists generated from both sets of scores were compared using Spearman’s rank correlation. GPT-4o was then used to create ten generic application files. These were scored by GAI and compared to GAI scores for the 2023-2024 CaRMS applicants using the Wilcoxon rank-sum test. Results: A total of 124 application files were included. Median GAI file review scores were consistently higher than RSC-assigned scores (24.46 vs. 17.54 y, p<0.05) and had less variance between applicants (6.96 vs. 20.80, p<0.05). The interrater reliability between GAI scores and RSC scores was poor across all application elements (0.16). Rank lists generated by GAI and RSC scores demonstrated a weakly positive correlation for each application element (0.25 to 0.37, p<0.05). Rank lists based on total file review scores demonstrated a moderately positive correlation (0.44, p<0.05). Median scores for GAI-created files compared to CaRMS applicant files were statistically similar for application CVs (6.88, p=0.25), but were significantly higher for other application elements and global scores (27.51 vs. 24.46, p<0.05). Conclusions: GAI in its current form cannot reliably replicate human file review. Further research is needed to determine the potential role for GAI in residency selection.

• AI Literacy and Training Needs in Midwifery Education: A National Mixed-Methods Study in France.

  Date Submitted: Mar 2, 2026

  Open Peer Review Period: Mar 6, 2026 - May 1, 2026

  • This manuscript needs more reviewers Peer Review Me

  Background: Artificial intelligence (AI) is increasingly shaping health care, yet the AI preparedness of midwifery students remains underdocumented. Evidence is needed to inform midwifery-specific curriculum development and to clarify how students understand and operationalize AI in training and placements. Objective: This mixed-methods study aimed to assess French midwifery students’ AI readiness, training needs, and ethical/regulatory concerns. Methods: We conducted a national sequential explanatory mixed-methods study during the 2024–2025 academic year. A web-based survey (five previously translated/adapted questionnaires) was disseminated via midwifery schools/universities in France (30/33, 91%, institutions confirmed dissemination and responses were received from these 30 institutions). Eligible participants were students enrolled in years 2–5 of the French midwifery curriculum. We computed mean theme scores (1–5) with 95% confidence intervals (CIs) and assessed internal consistency using Cronbach α. Analyses were restricted to fully completed questionnaires. Semi-structured interviews were conducted with volunteer students from one midwifery school in Eastern France (n=8), transcribed verbatim, anonymized, and analyzed using thematic analysis. Mixed-methods integration used a joint display. Results: Of 414 survey entries, 190 were fully completed and kept for analysis (190/414, 46.1%). Mean theme scores for AI skills and knowledge were below the neutral midpoint, ranging from 1.20 (95% CI 1.09–1.23) for familiarity with advanced AI techniques to 2.89 (95% CI 2.48–3.31) for analytical concepts in AI for health. Perceived ability to use AI for clinical purposes was low (2.05, 95% CI 1.01–3.09). In contrast, students strongly endorsed AI education (belief that students and professionals should be trained: 4.11, 95% CI 3.96–4.26) and emphasized evidence and safety requirements (up to 4.04, 95% CI 3.84–4.24). Item-level results suggested “AI label ambiguity”: general AI familiarity showed higher agreement (91/190, 47.9%) than familiarity with specific concepts such as machine learning (20/190, 10.5%) or deep learning (13/190, 6.8%). Interviews aligned with these patterns, indicating rare exposure to explicitly identified AI-supported workflows in placements and describing mainly academic and informal uses of generative tools. Participants emphasized patient safety, accountability, and preservation of human judgment. Conclusions: French midwifery students report a substantial AI readiness gap characterized by both low technical preparedness and limited situated exposure during placements, despite strong demand for training and high salience of safety and governance. Findings support implementing a structured, progressive curriculum linked to midwifery-relevant clinical scenarios and aligned with placement ecosystems. Future measurement should explicitly distinguish generative AI practices from regulated clinical AI systems and capture safe-use behaviors to improve construct validity.

• Automated Discovery and Application of Surgical Feedback Quality Criteria Using Large Language Models (LLMs): AI-Based Evaluation Study

  Date Submitted: Mar 5, 2026

  Open Peer Review Period: Mar 5, 2026 - Apr 30, 2026

  • Peer Review Me

  Background: Verbal feedback delivered by attending surgeons in the operating room plays a critical formative role in resident trainee skill acquisition. Yet, assessing the quality of trainer feedback and its effectiveness in influencing trainee behavior during live surgery remains a challenge. Prior studies relied on extensive manual annotation by expert human raters and focused on broad taxonomies that overlook the qualitative aspects of feedback delivery such as clarity or urgency. Limited existing automated methods, including keyword analysis and topic modeling, also fail to capture these nuanced delivery dimensions. Objective: The study aimed to develop and evaluate a scalable, automated framework for discovering and scoring interpretable surgical feedback quality criteria grounded in real-world surgical training interactions and clinically validated outcome measures. Methods: We introduce a two-stage large language model (LLM)-based framework. In the first stage, multi-agent prompting with multiple GPT-4o instances, seeded with clinically validated definitions of feedback effectiveness outcomes and unlabeled feedback examples, independently proposes candidate quality criteria. These are consolidated via hierarchical clustering and a deterministic LLM synthesis step into six human-interpretable, behaviorally anchored dimensions: Encouraging, Urgent, Actionable, Timely, Clear, and Reflective. In the second stage, these criteria are applied to score feedback instances at scale using an LLM-as-a-judge approach. Framework evaluation included predictive modeling of four clinically annotated behavioral outcomes, statistical significance testing using DeLong's method, generalized linear mixed modeling of associations between quality dimensions and outcomes, and human-AI alignment assessment using quadratically weighted Cohen's kappa on a stratified sample of instances rated by two domain-expert human raters. Results: Applied to 4,210 intraoperative feedback instances, the six AI-discovered quality criteria achieved AUROCs of 0.75 (95% CI: 0.74–0.77) for trainee behavior change and 0.71 (95% CI: 0.69–0.72) for trainee verbal response. Augmenting prior automated topic modeling features with our criteria yielded consistent gains of 9–12% across all four behavioral outcomes. DeLong's testing confirmed statistically significant additive predictive value of the AI-derived dimensions over both topic modeling and human-annotated baselines. Generalized linear mixed modeling revealed that Actionable (rate ratio [RR]=1.22), Timely (RR=1.24), and Urgent (RR=1.11) feedback were significantly associated with trainee behavioral adjustment, while Reflective (RR=1.34) and Clear (RR=1.13) feedback predicted verbal acknowledgment. Human-AI alignment was substantial for five of six dimensions (quadratically weighted κ=0.60–0.79), approaching inter-human agreement levels. Conclusions: Our LLM-based framework enables scalable, interpretable, and clinically grounded assessment of surgical feedback delivery quality, without requiring manual annotation. The discovered criteria demonstrate significant predictive validity for real-world trainee and trainer behavioral outcomes and exhibit strong alignment with expert human judgment, providing a foundation for improving intraoperative teaching and surgical education quality assurance.

• Factors Associated with Engagement When Using Social Media to Teach Point-of-Care Ultrasound in Internal Medicine Residency

  Date Submitted: Mar 2, 2026

  Open Peer Review Period: Mar 4, 2026 - Apr 29, 2026

  • This manuscript needs more reviewers Peer Review Me

  Background: Although social media is often viewed by residents and could be used to reinforce teaching points, there is little data on methods that improve engagement in learning medical topics through this medium. Objective: We observed how the timing of posted questions, answering questions correctly, and giving supportive comments affected the engagement of residents learning point-of-care ultrasound on social media. Methods: Of 60 medical residents, 35 followed an Instagram account that posted ultrasound video clips with questions during the academic year. Engagement, E, was the percentage of questions answered of the total number of clips viewed for each post and each resident. E was tested for an association with (1) weekend vs. weekday posts, (2) answering questions correctly vs. incorrectly, and (3) supportive responses from faculty vs. no feedback. Results: Of 16 posts, 120 questions were answered from 428 clips viewed by 32 residents, for an E =28% [range: 15-59%] for posts and a median (IQR) E=19% (0-39%) for residents with 71% (n=25) engaging on at least one post. E was higher during weekdays vs. weekends, 30% vs. 21% (p=0.007), and correlated to answering correctly vs. incorrectly (r=0.6, p<0.001). A supportive comment resulted in a lower percentage of answering the next post, compared to no feedback (30% vs. 71%, p=0.02). Conclusions: Resident engagement in social media was higher with having questions answered correctly, but, surprisingly, was lower when posting during weekends and immediately after receipt of a supportive comment.

• Bridging Inner Growth and Global Health: A Position Paper on Embedding IDGs in Health Professions Curricula

  Date Submitted: Feb 25, 2026

  Open Peer Review Period: Feb 26, 2026 - Apr 23, 2026

  • Peer Review Me

  Global transformations, including demographic aging, climate-related health risks, and rapid technological acceleration are reshaping health systems and the competencies required of future healthcare professionals. Yet current curricula often struggle to integrate these complex challenges in a coherent and future-oriented manner. This Eye Opener highlights the potential of the Inner Development Goals (IDGs) as an underutilized conceptual framework for enriching competency-based education in the health professions. The IDGs emphasize five dimensions: Being, Thinking, Relating, Collaborating, and Acting that align with key professional capacities such as self-awareness, systems thinking, empathy, interprofessional teamwork, and ethical action. Drawing on examples from geriatric care, climate-adapted practice, and AI-supported clinical reasoning, we illustrate how IDG-aligned learning outcomes can complement existing competency frameworks by fostering inner capacities essential for clinical judgement and person centered care. At the same time, we provide a critical reflection on potential risks, including over individualization of responsibility, insufficient attention to structural determinants of health, and tensions with assessment-driven educational cultures. Rather than proposing IDGs as a complete solution, this article argues that they offer a valuable conceptual entry point for rethinking how health professions education can prepare learners for the uncertainties, ethical complexities, and interdependencies of contemporary healthcare. The IDGs can help open new pedagogical and conceptual spaces, encouraging educators to design learning environments that support both technical proficiency and the inner capacities needed for navigating an increasingly complex world.

• Smells Like Real-World Spirit – Olfactory Cues and Patient Safety Behavior in VR Medical Simulation: a Randomized Controlled Trial

  Date Submitted: Feb 23, 2026

  Open Peer Review Period: Feb 24, 2026 - Apr 21, 2026

  • Peer Review Me

  Background: Simulation-based medical education is essential for improving patient safety. In virtual reality (VR)–based simulation, immersion is primarily generated through visual and auditory cues, while other sensory modalities are typically absent. This sensory limitation may reduce the emergence of authentic safety-relevant behaviors. Olfaction plays an important role in clinical reasoning, risk perception, and self-protective behavior and is closely linked to memory and emotion. Although olfactory cues have been shown to influence hand hygiene behavior in real or simulated-real environments, their targeted integration into fully immersive VR-based medical simulation has not been systematically examined. Objective: This study aimed to investigate whether adding a real olfactory cue (disinfectant scent) to a fully virtual clinical simulation increases patient safety–relevant behavior, specifically hand hygiene compliance (hand disinfection and glove usage). Methods: In a randomized controlled study at the University of Münster (winter term 2025/26), 89 medical students participated in a VR-based clinical simulation. Study rooms were pre-assigned to either an olfactory intervention or a control condition, and participants selected their room without knowledge of the assigned condition. Hand hygiene and glove use were automatically tracked as outcomes. Odds ratios were calculated to assess the effect of the intervention on these behaviors. Results: The olfactory intervention nearly tripled the odds of hand disinfection (OR = 2.81, 95% CI 1.09–7.75, P = 0.037), while no significant difference was observed for glove use (OR = 1.62, P = 0.278). Conclusions: The integration of a real olfactory cue into a fully immersive VR medical simulation significantly increased hand disinfection behavior, particularly after patient contact, but did not affect glove use. These findings suggest that olfactory augmentation can selectively reinforce safety-relevant behaviors in digital training environments. Incorporating real-world sensory cues into VR may represent a simple yet effective design strategy to enhance behavioral authenticity and patient safety outcomes in simulation-based medical education. Clinical Trial: German Clinical Trials Register: DRKS00039472