JMIR Medical Education最新文献

Background: Clinicians are central to treating tobacco use disorder, yet practical training is inconsistent, and confidence varies. Brief, text message-based microlearning may offer a low-burden way to strengthen foundational competencies in busy clinical settings.

Objective: This paper aims to evaluate whether a short SMS microlearning series improves clinicians' self-reported confidence in managing tobacco use disorder.

Methods: We conducted a single-arm, pre-post educational pilot at an academic medical center. A brief formative survey (13 items; 106 respondents) identified local knowledge gaps and informed message topics and sequencing. The 13-day series delivered 1 concise message per day with key teaching points and links to curated resources. The prespecified primary outcome was self-reported confidence in managing tobacco use disorder (1-100 scale) measured immediately before and after the series. Of the 34 clinicians who signed up, 22 completed the baseline questionnaire and enrolled (attendings: n=4, 18%; trainees: n=18, 82%). Changes in confidence among participants with paired ratings were tested with a paired t test. Engagement with embedded links was recorded.

Results: All enrolled participants completed the 13-day series; none unsubscribed. Postintervention confidence ratings were provided by 18 participants. Mean confidence increased from 60 (SD 16) at baseline to 85 (SD 10) after the series (t17=-10.71; P<.001). Embedded links were opened in 67% (178/266) of messages. Free-text feedback was predominantly positive and emphasized the convenience, clarity, and point-of-care usefulness of brief messages.

Conclusions: A brief SMS microlearning series was associated with a substantial improvement in clinicians' confidence to manage tobacco use disorder, with high completion and evidence of engagement. This low-cost, scalable approach appears practical for busy clinicians. Findings should be interpreted cautiously given the single-arm design, self-selection, and reliance on self-reported confidence rather than objective knowledge or clinical outcomes. Future studies should include a validated knowledge assessment, a randomized comparison, broader sampling, and follow-up to assess durability and impact on care.

Background: Medical education continues to favor didactic lectures as the predominant method of instruction. However, in recent years, there has been a shift toward active learning methodologies such as gamification.

Objective: This study aimed to describe the implementation of 3 open-access, web-based pharmacology games tailored for medical students: Cross DRUGs, Find the DRUG, and DRUGs Escape Room. The study also evaluated the impact of gamification on knowledge retention, student engagement, and learning experience in pharmacology education.

Methods: We used a quasi-experimental design to examine the effects of gamification on knowledge retention by comparing pretest and posttest scores between the gamer and control groups. Each week, students self-selected into either the gamer group or the control group based on personal preference. All students were provided with online access to the same lecture slides. Students in the control group completed both the pretest and posttest but did not play any of the games. A survey was administered to assess students' perceptions of gamification as a learning tool.

Results: Of the 72 students enrolled in the course, 49 (68%) agreed to participate, with 40 (56%) students completing both the pretest and posttest and being included in our analysis. As participation could vary weekly, an individual student might have appeared in both groups across different weeks, resulting in 59 gamer sessions and 20 control sessions. The mean pretest scores were 6.05 (SD 2.31) for the control group and 6.20 (SD 2.13) for the gamer group. The mean posttest scores were 6.90 (SD 2.02) for the control group and 8.47 (SD 1.30) for the gamer group. The gamer group exhibited significantly improved posttest scores (P=.006), while the control group did not (P=.21). Most respondents (25/30, 83%) found the games enjoyable and agreed that the games effectively helped them understand pharmacological concepts (24/30, 80%). Additionally, 70% (21/30) of students believed they learned better from the gaming format than from didactic lectures. Most favored a blended approach that combines lectures with games or case studies.

Conclusions: Gamification can serve as an effective complementary teaching tool for helping medical students learn pharmacological concepts.

Unlabelled: Artificial intelligence (AI) has the potential to transform medical training through adaptive learning, immersive simulations, automated assessments, and data-driven insights, offering solutions to persistent issues such as high student-to-faculty ratios, overcrowded classrooms, and limited clinical exposure. Globally, many universities have already embedded AI literacy and competencies into undergraduate, postgraduate, and continuing education programs, while in Vietnam, the use of AI in medical education remains limited and fragmented. Most students have little formal exposure to AI, and empirical evidence on faculty or institutional readiness is scarce. Experiences from other countries, including Malaysia, Palestine, and Oman, demonstrate that incremental adoption and faculty development can facilitate cultural acceptance and curricular innovation, providing useful lessons for Vietnam. At the same time, significant barriers remain. These include inadequate infrastructure in provincial universities, low levels of AI literacy among both students and educators, underdeveloped regulatory and ethical frameworks, and resistance to pedagogical change. Cost-effectiveness and sustainability are additional concerns in a middle-income context, where upfront investments must be balanced against long-term benefits and equitable access. Advancing AI in Vietnamese medical education will therefore require a coordinated national strategy that prioritizes infrastructure, AI literacy, faculty development, quality assurance, and sustainable funding models, alongside ethical and legal safeguards. By addressing these key foundations, Vietnam can harness AI not only to modernize medical education but also to strengthen preparedness for a digitally enabled health workforce.

Background: The impact of Pass/Fail or Tiered grade assessment for exams in undergraduate medical education has caused much debate, but there is little data to inform decision-making. The increasing number of medical schools transitioned to a Pass/Fail assessment has raised concerns about medical students' academic performance. In 2018, during the undergraduate medical curriculum reform at the Faculty of Medicine, Aalborg University changed some exams from Pass/Fail to Tiered grade and vice versa for other exams. These changes provide an opportunity to evaluate the different assessment forms.

Objective: This study aimed to evaluate medical students' academic performance at the final licensing exam in relation to the exam grading principle.

Methods: This single-center cohort study at Aalborg University Medical School, North Denmark Region, assesses the change from 2-digit Tiered grade to Pass/Fail evaluation and vice versa of undergraduate medical students' exams after the 4th and 5th year clinical training modules from Autumn 2015 through Spring 2023. The primary outcome was (1) the average grades at the final licensing exam and (2) the number of students failing exams during the previous two years.

Results: Among the total of 7634 exams, 7164 4th and 5th year clinical training exams were included in the comparisons, of which 3047 (42.5%) were Pass/Fail exams and 4117 (57.5%) were Tiered grade exams. The frequency of students failing exams was 3.3% (n=101/3047) at Pass/Fail and 1.97% (81/4117) with Tiered grade exams (P<.001). This difference was leveled out when counting the near-failure tiered grade as Fail. Tiered grade exams did not differ between semesters (P=.99) nor show a time trend at the 4th year (P=.66). The final licensing exam grades were unaltered (P=.47).

Conclusions: Contrary to our expectation, Pass/Fail exams exhibited a higher fail rate compared to Tiered grade exams without lowering the final academic performance. These results suggest that a shift from Tiered grades to Pass/Fail assessment redirects the focus from rewarding high performance to ensuring standards are maintained among underperforming students.

Background: Large language models (LLMs) offer the potential to improve virtual patient-physician communication and reduce health care professionals' workload. However, limitations in accuracy, outdated knowledge, and safety issues restrict their effective use in real clinical settings. Addressing these challenges is crucial for making LLMs a reliable health care tool.

Objective: This study aimed to evaluate the efficacy of Med-RISE, an information retrieval and augmentation tool, in comparison with baseline LLMs, focusing on enhancing accuracy and safety in medical question answering across diverse clinical domains.

Methods: This comparative study introduces Med-RISE, an enhanced version of a retrieval-augmented generation framework specifically designed to improve question-answering performance across wide-ranging medical domains and diverse disciplines. Med-RISE consists of 4 key steps: query rewriting, information retrieval (providing local and real-time retrieval), summarization, and execution (a fact and safety filter before output). This study integrated Med-RISE with 4 LLMs (GPT-3.5, GPT-4, Vicuna-13B, and ChatGLM-6B) and assessed their performance on 4 multiple-choice medical question datasets: MedQA (US Medical Licensing Examination), PubMedQA (original and revised versions), MedMCQA, and EYE500. Primary outcome measures included answer accuracy and hallucination rates, with hallucinations categorized into factuality (inaccurate information) or faithfulness (inconsistency with instructions) types. This study was conducted between March 2024 and August 2024.

Results: The integration of Med-RISE with each LLM led to a substantial increase in accuracy, with improvements ranging from 9.8% to 16.3% (mean 13%, SD 2.3%) across the 4 datasets. The enhanced accuracy rates were 16.3%, 12.9%, 13%, and 9.8% for GPT-3.5, GPT-4, Vicuna-13B, and ChatGLM-6B, respectively. In addition, Med-RISE effectively reduced hallucinations, with reductions ranging from 11.8% to 18% (mean 15.1%, SD 2.8%), factuality hallucinations decreasing by 13.5%, and faithfulness hallucinations decreasing by 5.8%. The hallucination rate reductions were 17.7%, 12.8%, 18%, and 11.8% for GPT-3.5, GPT-4, Vicuna-13B, and ChatGLM-6B, respectively.

Conclusions: The Med-RISE framework significantly improves the accuracy and reduces the hallucinations of LLMs in medical question answering across benchmark datasets. By providing local and real-time information retrieval and fact and safety filtering, Med-RISE enhances the reliability and interpretability of LLMs in the medical domain, offering a promising tool for clinical practice and decision support.

Background: Clinical reasoning is increasingly recognized as an important skill in the diagnosis of common and serious conditions. eCREST (electronic Clinical Reasoning Educational Simulation Tool), a clinical reasoning learning resource, was developed to support medical students to learn clinical reasoning. However, primary care teams now encompass a wider range of professional groups, such as physician assistants (PAs), who also need to develop clinical reasoning during their training. Understanding PAs' clinical reasoning processes is key to judging the transferability of learning resources initially targeted to medical students.

Objective: This exploratory study aimed to measure the processes of clinical reasoning undertaken on eCREST by PA students and compare PAs' reasoning processes with previous data collected on medical students.

Methods: Between 2017 and 2021, PA students and medical students used eCREST to learn clinical reasoning skills in an experimental or learning context. Students undertook 2 simulated cases of patients presenting with lung symptoms. They could ask questions, order bedside tests, and select physical exams during the case to help them form, reflect on, and reconsider diagnostic ideas and management strategies while completing a case. Exploratory analysis was undertaken by comparing students' data gathering, flexibility in diagnosis, and diagnostic ideas between medical and PA students.

Results: In total, 159 medical students and 54 PA students completed the cases. PAs were older (mean 27, SD 7 y vs mean 24, SD 4 y; P<.001) and more likely to be female (43/54, 80% vs 84/159, 53%; P<.001). Medical and PA students were similar in the proportion of essential questions asked (Case 1: mean 70.1 vs mean 73.2; P=.33; Case 2: mean 74.6 vs mean 70.9; P=.27), physical examinations requested (Case 1: mean 54.7 vs mean 54.0; P=.59; Case 2: mean 69.3 vs mean 67.5; P=.59), bedside tests selected (Case 1: mean 74.4 vs mean 83.3; P=.05; Case 2: mean 47.9 vs mean 50.0; P=.69), and number of times they changed their diagnoses (Case 1: mean 2.8 vs mean 2.8; P=.99; Case 2: mean 2.8 vs mean 2.5; P=.81). Both student groups improved in their diagnostic accuracy during the cases.

Conclusions: These results provide suggestive evidence that medical and PA students had similar clinical reasoning styles when using an online training tool to support their diagnostic decision-making.