Curriculum-Aligned Conversational AI to Support Pathology Self-Directed Learning in Preclinical Medical Students: Crossover Randomized Pilot Study.
Kasie Roark, Yumin Choi, Elizabeth Hoy et al.
✦ AI-curated · Sources linked
In 30 seconds
This crossover randomized pilot study evaluated a curriculum-aligned custom GPT for self-directed pathology learning among 44 second-year medical students at the University of South Carolina, Floyd School of Medicine. Students used the GPT to prepare for one faculty-led case discussion and conventional resources for another. Compared with conventional preparation, GPT-supported preparation was associated with higher self-directed learning rubric scores (10.3 vs 9.4 out of 12; mean difference 0.9, 95% CI 0.5-1.3; P<.001) and discussion scores. Delayed retention did not differ significantly in the primary analysis.
Key findings
- Self-directed learning scores were higher with the custom GPT: 10.3 vs 9.4 out of 12; mean difference 0.9, 95% CI 0.5-1.3; P<.001.
- Small group discussion scores were higher with GPT-supported preparation: 14.9 vs 13.7 out of 16; mean difference 1.2, 95% CI 0.7-1.7; P<.001.
- Delayed retention on end-of-block examination items did not differ significantly (Wilcoxon signed-rank P=.052).
- Of 44 students, 34 (77%) reported clearer reasoning and 31 (70%) reported improved conceptual understanding.
Why it matters
The findings suggest that a curriculum-aligned conversational AI tool with structured Socratic prompts may support preparation and participation in pathology discussions more effectively than unguided conventional preparation. The absence of a significant retention difference cautions against assuming that improved learning-process or discussion scores translate into better delayed knowledge. Larger studies can clarify the educational value and generalizability of this approach.
What to keep in mind
This was a pilot study involving 44 voluntary participants at one medical school and two pathology cases. Delayed retention did not differ significantly, and the findings may not generalize to other curricula or settings.