Number
801
Name
Enhancing Clinical Reasoning in Second-Year Medical Students Through an AI-Augmented Self-Directed Learning Experience: A Pilot Study
Date & Time
Monday, June 8, 2026, 6:00 PM - 7:30 PM
Location Name
Oglethorpe Ballroom
Speakers
Authors
Ioulia Chatzistamou MD, PhD1, 1Department of Pathology, Microbiology & Immunology, University of South Carolina School of Medicine, Columbia
Kasie Roark MS, PhDc1,2,3, 1Department of Pathology, Microbiology & Immunology, University of South Carolina School of Medicine, Columbia, 2Office of Research, University of South Carolina, Columbia, 3AI in Medicine Extracurricular Track, University of South Carolina School of Medicine, Columbia
Yumin Choi BS3, 3MD Candidate, Class 2028, AI in Medicine Extracurricular Track, University of South Carolina School of Medicine, Columbia
Sarah McMillan BS3, 3MD Candidate, Class 2028, AI in Medicine Extracurricular Track, University of South Carolina School of Medicine, Columbia
Noel Tufts MPH3, 3MD Candidate, Class 2028, AI in Medicine Extracurricular Track, University of South Carolina School of Medicine, Columbia
Sweta Chalise MPH, CPH2, 2Office of Research, University of South Carolina, Columbia
Brian Keisler, MD4, 4Department of Family and Preventive Medicine, University of South Carolina School of Medicine, Columbia
Presentation Topic(s)
Technology and Innovation
Description
PURPOSE
Artificial intelligence (AI) is opening new opportunities to enrich
diagnostic reasoning and self-directed learning (SDL) in undergraduate
medical education. At the University of South Carolina School of Medicine,
Columbia (USCSOM-C), second-year medical students complete SDL exercises as
preparation for pathology small-group discussions in the
Endocrine/Reproductive System Block. To enhance these experiences, we
developed a customized GPT-based AI assistant designed to support students’
clinical reasoning by prompting reflection, suggesting credible resources,
and guiding their inquiry through complex pathology cases.
METHODS
In this randomized crossover pilot study, students will alternate between
an AI-Augmented Group, using the GPT-based reasoning assistant, and a
Traditional Group, using conventional resources (PowerPoint slides,
textbooks, and articles) without ChatGPT access. Two small group discussions
will be held on Cushing’s disease and prostate cancer cases. Students will
complete a structured SDL worksheet and a REDCap reflection survey evaluating
their perceptions of AI’s educational value. Faculty will assess SDL
documentation and group performance using validated LCME-aligned rubrics.
Quantitative data (rubric scores) will be analyzed using paired statistical
tests, while qualitative reflections will undergo thematic analysis to
explore reasoning depth and metacognitive engagement.
RESULTS
The GPT assistant and survey have been developed and pilot tested. The
student cohort has been identified and stratified, and faculty raters have
received training. Implementation is scheduled for the early spring
Endocrine/Reproductive Block, with data collection during SDL and case-based group
sessions.
CONCLUSION
We expect that AI-augmented SDL will promote deeper diagnostic reasoning,
more intentional resource selection, and richer reflection compared to
traditional methods. This pilot aims to inform the responsible integration of
generative AI into a case-based approach and provide evidence for its
potential to enhance clinical reasoning development in pre-clerkship medical
education.
Artificial intelligence (AI) is opening new opportunities to enrich
diagnostic reasoning and self-directed learning (SDL) in undergraduate
medical education. At the University of South Carolina School of Medicine,
Columbia (USCSOM-C), second-year medical students complete SDL exercises as
preparation for pathology small-group discussions in the
Endocrine/Reproductive System Block. To enhance these experiences, we
developed a customized GPT-based AI assistant designed to support students’
clinical reasoning by prompting reflection, suggesting credible resources,
and guiding their inquiry through complex pathology cases.
METHODS
In this randomized crossover pilot study, students will alternate between
an AI-Augmented Group, using the GPT-based reasoning assistant, and a
Traditional Group, using conventional resources (PowerPoint slides,
textbooks, and articles) without ChatGPT access. Two small group discussions
will be held on Cushing’s disease and prostate cancer cases. Students will
complete a structured SDL worksheet and a REDCap reflection survey evaluating
their perceptions of AI’s educational value. Faculty will assess SDL
documentation and group performance using validated LCME-aligned rubrics.
Quantitative data (rubric scores) will be analyzed using paired statistical
tests, while qualitative reflections will undergo thematic analysis to
explore reasoning depth and metacognitive engagement.
RESULTS
The GPT assistant and survey have been developed and pilot tested. The
student cohort has been identified and stratified, and faculty raters have
received training. Implementation is scheduled for the early spring
Endocrine/Reproductive Block, with data collection during SDL and case-based group
sessions.
CONCLUSION
We expect that AI-augmented SDL will promote deeper diagnostic reasoning,
more intentional resource selection, and richer reflection compared to
traditional methods. This pilot aims to inform the responsible integration of
generative AI into a case-based approach and provide evidence for its
potential to enhance clinical reasoning development in pre-clerkship medical
education.