Clinical Research Directory

AI-Generated and Rule-Based Individualized Feedback for Medical Students' Nephrology Learning

Medical education increasingly emphasizes individualized learning and formative assessment; however, providing timely and personalized feedback to students remains challenging because of limited faculty resources. Generative artificial intelligence (AI) may provide a scalable approach for generating individualized feedback based on students' learning needs, but its educational effectiveness compared with traditional feedback approaches remains uncertain. This study is a single-center, prospective, three-arm, cluster randomized controlled trial designed to evaluate the effect of pretest-based individualized feedback on medical students' learning outcomes during a nephrology clerkship. Undergraduate medical students participating in a nephrology clerkship at The Second Qilu Hospital of Shandong University will be recruited and assigned by teaching class clusters to one of three groups: an AI-generated feedback group, a rule-based individualized feedback group, or a no-feedback control group. All participants will receive the same nephrology clerkship teaching based on the BOPPPS-CBL instructional model. Before the clerkship, students will complete a diagnostic pretest. Based on the pretest results, individualized feedback will be provided according to group allocation. The primary outcome is immediate knowledge acquisition assessed by the post-test score immediately after the clerkship. Secondary outcomes include four-week knowledge retention, knowledge gap remediation, and clinical reasoning performance. This study aims to determine whether AI-generated individualized feedback provides additional educational benefits compared with traditional rule-based feedback and no feedback, and to provide evidence for the application of artificial intelligence-assisted personalized learning in medical education.

Intelligent Simulated Patient System for Gastroenterology Education

This randomized controlled trial evaluated a large language model-based Intelligent Simulated Patient System (ISPS) for gastroenterology history-taking education. Ninety medical students were randomly assigned in a 1:1 ratio to 4 weeks of conventional clinical learning or ISPS-assisted training. The ISPS used teacher-defined structured case information to support simulated patient interactions and generated item-level formative feedback using predefined scoring rubrics. The implemented system did not use external electronic health record retrieval, retrieval-augmented generation, a vector database, or semantic-similarity threshold scoring. The primary outcomes were post-intervention medical history-taking performance and clinical diagnostic accuracy. History-taking performance was assessed in a standardized Objective Structured Clinical Examination by two senior physicians who were blinded to group allocation. Secondary outcomes included patient-centered communication competence and participants' acceptance of the ISPS.

Teaching Doctors in Training to Reason With Artificial Intelligence

The purpose of the TEACH-AI study is to assess whether a brief, structured workshop on artificial intelligence can improve the performance of medicine doctors in training (i.e. residents) in their diagnostic and management reasoning. In this multi-site randomized controlled trial, internal medicine and family medicine residents are assigned either to receive an in-person workshop on safe, effective LLM use before a standardized AI-assisted assessment, or to complete the same assessment before receiving the workshop. Residents will review clinical cases that are fully synthetic, no protected health information is used, using a password protected LLM interface.

Emergency Structured Clinical Examination Training bOX

This study assessed whether a training toolbox called ECOX could help medical students prepare for their objective structured clinical examination (OSCE) during an emergency medicine clerkship. The hypothesis was that adding ECOX to usual training would improve examination performance and reduce examination-related anxiety. Eight 10-week clerkship rotations at Angers University Hospital, France, were randomly assigned to usual training alone or usual training plus ECOX, with four rotations in each group. All students in a rotation received the same assignment. ECOX provided a timer and reference assessment grids for practising seven-minute encounters with real patients who agreed to participate. Students could use the toolbox throughout their clerkship. Both groups completed the same three-station examination at the end of the clerkship. Examiners were unaware of group allocation. Anxiety was assessed at the beginning of the clerkship and immediately before the final examination.

Telemedicine Education Program for Medical Students

This study evaluated a structured telemedicine education program for undergraduate medical students. Third-year medical students participated in either a structured telemedicine education program or a control condition. The education program included theoretical sessions, observation of real telemedicine consultations, simulated patient encounters, and structured feedback. Students' telemedicine knowledge was assessed before and after the study, and their telemedicine consultation skills were evaluated through a standardized simulated patient encounter. The study aimed to examine whether structured telemedicine education could improve medical students' knowledge and consultation skills.

Effectiveness of AI-generated Flashcards for Enhancing Learning Outcomes in Undergraduate Biochemistry

This study evaluated the effectiveness of artificial intelligence (AI)-generated flashcards in improving learning outcomes among undergraduate medical students studying biochemistry. AI-generated flashcards were developed using generative AI tools and incorporated into teaching sessions on selected biochemistry topics. A prospective non-randomized crossover study was conducted among first-year Bachelor of Medicine and Bachelor of Surgery (MBBS) students at KMSK Government Medical College, Chandrapur, India. Participants received instruction through both AI-generated flashcards and conventional teaching methods in different study periods. Learning outcomes were assessed using pre-tests, post-tests, and delayed retention tests. Student perceptions regarding the usefulness, engagement, and acceptability of AI-generated flashcards were also evaluated. The study aimed to determine whether AI-generated flashcards could enhance immediate knowledge acquisition and long-term retention compared with conventional teaching methods in undergraduate biochemistry education.

Virtual Reality for Teaching Abscess Incision and Drainage to Medical Students

This completed randomized educational trial evaluated whether virtual reality (VR)-assisted teaching improves medical students' performance in abscess incision and drainage compared with conventional teaching. A total of 120 second-year medical students were randomly assigned in a 1:1 ratio to either VR-assisted teaching or conventional teaching. Both groups received 120 minutes of instruction covering the same core learning objectives. The primary outcome was the immediate post-intervention objective structured clinical examination (OSCE) skills score for abscess incision and drainage. Secondary outcomes included theoretical knowledge, learning interest, and course satisfaction.

Digital Microlearning for Patient-Safety Readiness in Nursing Students

This randomized controlled trial evaluated whether a seven-day, patient-safety-focused digital microlearning program improved sustained clinical error recognition and the transfer of safety learning among second-year nursing students entering surgical clinical practice. Ninety students were randomly assigned in a 1:1 ratio to digital microlearning plus standard education or standard education alone. Outcomes were assessed at baseline, immediately after the intervention, at the end of the first week of surgical clinical practice, and at the end of the seventh week of surgical clinical practice. Weekly ecological momentary assessment prompts examined safety-oriented behavior during clinical practice. Blinded performance tasks assessed objective clinical transfer, and an explanatory qualitative component explored how and under what conditions students applied the learning in practice.

Integrated PBL-CBL-CTTM Teaching to Enhance Clinical Self-Efficacy in Endocrine Critical Illness Among Internal Medicine Residents: An RCT

This randomized controlled trial evaluates whether an integrated online-offline multimodal teaching approach combining Problem-Based Learning (PBL), Case-Based Learning (CBL), and Case Three-Dimensional Teaching Method (CTTM) improves clinical self-efficacy and hands-on proficiency in managing endocrine and metabolic critical illness among internal medicine residents, compared to traditional lecture-based learning (LBL). Ninety internal medicine residents on their first endocrine rotation will be randomly assigned in a 1:1:1 ratio to three groups: (1) Control group receiving traditional LBL alone; (2) Experimental group 1 receiving PBL plus LBL; (3) Experimental group 2 receiving the integrated PBL-CBL-CTTM closed-loop teaching program. All groups receive a standardized 12-hour curriculum on endocrine and metabolic critical illness, including diabetic ketoacidosis, hyperosmolar hyperglycemic syndrome, thyroid storm, and refractory hypoglycemia. The primary outcome is the clinical skills assessment score (0-50 points). Secondary outcomes include theoretical examination scores, General Self-Efficacy Scale (GSES) scores, a self-developed Endocrine High-Risk Management Self-Efficacy Source Questionnaire (four dimensions based on Bandura's self-efficacy theory), teaching satisfaction, and semi-structured focus group interviews. The study was conducted from September 2025 to August 2026 at the Fourth Affiliated Hospital of Anhui Medical University, Chaohu, China. This study aims to establish an evidence-based, replicable teaching model that addresses the gap between theoretical knowledge and clinical confidence in managing endocrine emergencies among residents.

Gamification in Asynchronous Faculty Development

TEACH\_UP is a parallel-group randomized trial that evaluated whether adding a gamified design layer to a content-equivalent asynchronous faculty development course improved teaching self-efficacy, teaching orientation, engagement, course participation, and early teaching performance among medical educators at the University of Pécs Medical School. Participants were allocated to a gamified Moodle course or a standard Moodle course containing the same modules, learning objectives, assessments, and feedback. The gamified version added map-based progression, conditional release, experience points, badges, progress visibility, and a leaderboard. Outcomes included pre-post teaching self-efficacy and teaching orientation, post-course engagement, Moodle analytics, course performance, microteaching ratings, and a protocol-specified three-month follow-up survey.

An Agentic Co-educator System for Personalized Medical Education

With the rise of generative artificial intelligence and large language models, medical education is undergoing a significant transformation. Numerous studies have highlighted the limitations of traditional educational knowledge acquisition and the potential impact of artificial intelligence on medical education, resident training, and continuing education for clinical practitioners. However, there is a lack of real-world experiments on the effectiveness of AI-integrated education. Artificial intelligence can provide extensive educational resources and tools that are not limited by geographical location or language, thereby lowering the barrier to accessing high-quality medical education and promoting educational equity. Nevertheless, the performance of AI models trained by different medical institutions or healthcare systems may vary. To offer a more universal, accessible, high-quality, and interconnected educational journey. We have developed a virtual ophthalmology teacher, which developed based on foundational model and large language models. This model, named EyeTeacher provide comprehensive theoretical knowledge and clinical skills enhancement for untrained medical students. To verify the effectiveness of our EyeTeacher across different national ophthalmology education systems and languages, we plan to implement a randomized controlled trial. This trial will assess the clinical capabilities of all participants and explore the advantages and disadvantages of the system compared to traditional teaching methods.

Remote Guidance for eFAST Training Using a Head-Mounted Device

This study looks at the best way to remotely teach and guide doctors-in-training to perform a bedside ultrasound exam called eFAST (extended Focused Assessment with Sonography for Trauma), which is used to quickly check injured patients for internal bleeding or air around the lungs. The study takes place in a simulation setting, with no real patients. Resident physicians who have completed the basic ultrasound courses each watch the same short instructional video and then perform three eFAST examinations on the same healthy volunteer. Participants are randomly assigned to one of three groups: (1) the instructional video only, with no live help; (2) the video plus real-time audio guidance from a remote expert who can see the live ultrasound image; or (3) the video plus guidance through a head-mounted device that also lets the expert see the participant's own point of view. Afterwards, experts who were not involved in a given exam review the recorded ultrasound images, without knowing which group the participant was in, and rate the image quality and skill. The study also measures how long the exams take, the participants' workload, and the comfort and usability of the system. The goal is to learn whether real-time remote guidance, and in particular guidance through a head-mounted device, helps trainees produce better and faster eFAST examinations than an instructional video alone.

Evaluating in Person Curriculum for Educating Obstetrical Residents on Better Caring for 2 Spirit Metis Patients Using and AI Assisted OSCE.

Two-Spirit (2S) Métis individuals often encounter significant barriers and discrimination in healthcare settings. This multisite randomized controlled feasibility trial evaluates a novel artificial intelligence (AI)-assisted Objective Structured Clinical Examination (OSCE) designed to assess and improve Obstetrics and Gynecology (Ob/Gyn) residents' competencies in delivering culturally safe, trauma-informed, and identity-affirming care to 2S Métis patients. Up to 90 Ob/Gyn residents across three Canadian academic sites (University of Alberta, University of Manitoba, and University of Saskatchewan) will be randomized 1:1 to complete either a 2S Métis Academic Half Day paired with AI-assisted OSCE evaluation or standard institutional anti-racism training with AI-assisted OSCE evaluation. The primary focus is to evaluate study feasibility (recruitment, retention, and delivery fidelity) and resident performance on clinical communication, trauma-informed care, and cultural humility evaluated using a 2S Métis co-developed scoring rubric.

Blended Cardiac Rehabilitation Training for Cardiology Residents

This randomized controlled trial compares supervised hospital-based practice with matched online activities as part of a common cardiac rehabilitation curriculum for cardiology residents. Ninety-eight residents will be allocated 1:1 to a blended group or an online control group. Both groups will receive the same seven-module online curriculum and the same total scheduled learning time. The blended group will complete a matched block of in-person practice in the cardiology ward, outpatient clinic, and cardiopulmonary exercise testing laboratory, whereas the control group will complete matched-duration supplementary online activities. The primary outcome is the post-training 25-item cardiac rehabilitation knowledge score adjusted for the baseline score. Key secondary outcomes are Mini-CEX and procedure-specific DOPS scores assessed by assessors blinded to allocation.

Explainable AI in Medical Education: CerViD-MultiModal Framework Trial

This study evaluates whether Explainable Artificial Intelligence (XAI) explanations integrated into medical training improve AI literacy, reduce cognitive workload, and enhance learner trust compared to traditional lecture methods. Third-year medical students participated in a randomized controlled trial assessing the CerViD-MultiModal diagnostic framework during a neuroimaging diagnostic module focused on fornix atrophy in early and late mild cognitive impairment

Effects of a Cadaveric Workshop on Self-Reported Confidence in Ultrasound-Guided Upper Extremity Procedures

This prospective observational study evaluates the effects of a structured cadaveric workshop on physicians' self-reported confidence in performing upper extremity ultrasound-guided procedures. The workshop consisted of theoretical instruction and supervised hands-on cadaveric practice focusing on shoulder, elbow, and wrist procedures. Confidence levels were assessed before and after the workshop using standardized questionnaires. The study also examined the influence of prior ultrasound experience, procedural volume, and academic seniority on baseline confidence and post-workshop improvement.

Assessment of Hysteroscopy Skills With a New Assessment Tool

Objective To investigate validity of the hysteroscopy assessment tool (HYSAT) for assessment of competence in a clinical environment. Methods Novices and experienced gynecologists are recruited from three hospitals and observed and assessed while performing hysteroscopy. Performances are assessed using the HYSAT tool by two independent raters. Validity evidence is gathered in accordance with Messick's framework: validity evidence for content was ensured in previously published Delphi study, response process is ensured by standardization of written rater instructions. Internal structure is explored using Cronbach's alpha for internal consistency reliability; inter-rater reliability and test-retest reliability are calculated as Pearson's r independently across all ratings. Relationship to other variables is investigated by comparing performances of the participants in each group. Consequences evidence is explored by calculating a pass/fail standard using the contrasting groups' standard setting method.

AI-Generated Acne Vulgaris Learning Material Study

This comparative educational intervention study evaluated the educational effectiveness of artificial intelligence-generated patient information material on acne vulgaris among early-year medical students. In the first phase, acne vulgaris information texts were generated using different artificial intelligence models and evaluated by two dermatologists using quality and understandability assessment tools. In the second phase, the best-performing artificial intelligence-generated material was compared with standard human-generated material from the Turkish Dermatology Association and a control group. A total of 190 first- and second-year medical students at Erciyes University Faculty of Medicine participated in the study. Participants were divided into three groups: artificial intelligence-generated material group, Turkish Dermatology Association material group, and control group. After the educational materials were presented, participants completed a 12-question knowledge assessment questionnaire on acne vulgaris.

Impact of Concept Mapping in Case-Based Learning Among Medical Students

This prospective randomized study aims to compare the impact of three Case-Based Learning (CBL) approaches on knowledge acquisition, knowledge retention, and clinical reasoning development among third-year medical students during their obstetrics and gynecology rotation. Participants are randomly assigned to one of three groups: CBL without concept mapping, CBL with a teacher-constructed concept map, or CBL with a concept map co-constructed by students during the learning session. Knowledge acquisition and retention are assessed using pre-tests, immediate post-tests, and delayed post-tests. Clinical reasoning and knowledge organization are evaluated using a synthesis exercise. The study seeks to determine whether concept mapping and the modality of its integration into CBL enhance learning outcomes and clinical reasoning skills in undergraduate medical education.

AI-Assisted Structured Guidance for Narrative Review Writing

This randomized, parallel-group, assessor-blinded educational study will evaluate whether one session of AI-assisted structured guidance improves the quality of narrative literature reviews written by medical postgraduate students. A total of 102 students enrolled in an academic writing course will be randomly assigned to either an AI-assisted structured guidance group or a free AI use group. Both groups may use AI tools under academic integrity requirements and will complete a narrative literature review according to the same course requirements. The guidance group will receive one structured guidance session approximately two weeks before final submission. The session will be based on the six dimensions of the Scale for the Assessment of Narrative Review Articles (SANRA). The primary outcome is the final SANRA total score of the submitted narrative literature review.

Case-Based Learning for Bedside Lung Ultrasound Teaching in ICU

This single-center randomized controlled study aims to evaluate the effectiveness of Case-Based Learning (CBL) compared with traditional teaching in clinical training of bedside lung ultrasound (BLUE) for emergency medicine residents and medical students. The hypothesis is that CBL improves theoretical understanding, practical ultrasound skills, and clinical reasoning in emergency settings.

Artificial Intelligence-Generated vs Academician-Developed Multiple True/False Questions in Anesthesiology Education

This prospective observational study aims to evaluate the effectiveness and educational value of artificial intelligence (AI)-generated multiple true/false questions compared to those developed by experienced academicians in anesthesiology training. A total of 27 anesthesiology residents will be included in the study. Question sets consisting of 200 multiple true/false items will be created, with half generated by academicians and the other half generated using an artificial intelligence model (ChatGPT-based system). The questions will be based on standardized educational materials from the anesthesiology training curriculum. Participants will complete the test in a single session. Each correct answer will be scored as one point, and total scores will be calculated. In addition to test performance, item difficulty, discrimination indices, and test reliability will be analyzed. Furthermore, participants' perceptions regarding question quality will be evaluated. The study aims to determine whether AI-generated questions can provide a reliable and effective alternative to traditional question development methods in medical education and contribute to more objective and standardized assessment processes.

Transfusion Camp for Medical Students in Rwanda

Transfusion medicine is closely linked with safe surgery. Transfusion Camp is a multidisciplinary educational project aimed at improving transfusion medicine knowledge in physicians. The course has been piloted in Rwanda, showing improvement in participant knowledge and confidence, resulting in its recommended implementation into the medical school curriculum. This project aims to evaluate a multi-day Transfusion Camp course mandatory for graduating medical students and first-year interns practicing in district hospitals in Rwanda. Its implementation is evaluated through pre- and post-course testing, collecting participants' blood ordering practices following the course, and conducting participant interviews about barriers to safe transfusion medicine in their hospital.

Conversational AI Study

This is a prospective, collaborative, mixed-methods study which includes anesthesiology trainees receiving one active intervention - a conversational artificial intelligence (CAI) simulation. The goal is to explore the capability of AI to provide high-fidelity simulations that can ultimately improve how healthcare professionals handle difficult conversations such as disclosing medical errors or mistakes.