Purpose
Case-Based Learning (CBL) is an inquiry-based approach to medical education that emphasizes active engagement, critical thinking, and the application of foundational concepts. At the University of Texas at Tyler School of Medicine (UTTSOM), students worked in CBL groups to develop learning objectives (LOs) to learn foundational concepts. We analyzed the effectiveness of artificial intelligence (AI) in evaluating CBL group-generated LOs as a tool to assess foundational knowledge acquisition in a CBL curriculum.

Methods
In this student-driven continuous quality improvement study, the coded UTTSOM curricular LOs were compared to LOs generated by CBL group (n=5). These analyses were conducted either by manually cross-checking the LOs or using NotebookLM.

Results
Manual analysis revealed that an average of 65.5% of the LOs created by the CBL groups matched the curricular LOs. Areas of alignment primarily included topics such as pathophysiology and the mechanisms of disease and drug actions. Variations occurred when CBL groups either created a single LO to cover curricular LOs or did not address curricular LOs for topics not explicitly mentioned in the case. AI demonstrated proficiency in recognizing similar topics between curricular and CBL group LOs but struggled to detect nuanced differences. It tended to overlook topics unless explicitly and identically worded to the curricular LOs, limiting its utility in evaluating self-directed learning outcomes.

Conclusion
AI's utility in comparing curricular LOs with CBL group LOs may be limited. While successfully detecting similarities, AI struggles to discern nuances in CBL-generated LOs that differ in wording or structure from curricular LOs. This underscores the need for human interpretation to accurately evaluate the depth and breadth of self-directed learning in CBL. Future studies to enhance the data analyses using AI to reinforce foundational knowledge acquisition in CBL curriculum are warranted.