Learning Outcomes Following a Short-Term AI-Based Tutoring Program in Undergraduate Physical Therapy Students

Objective: This study aimed to develop and implement an artificial intelligence (AI) based cognitive-responsive tutor system for healthcare students and evaluate its effectiveness in regard to improving learning achievement through pre- and post-assessment analyses. Design: one-group pre–post exploratory design Methods: Fifty physical therapy students participated in an experimental study conducted at Kyungbok University using an AI-based cognitive-responsive tutor system that applied Deep Knowledge Tracing (DKT) and Bayesian Knowledge Tracing (BKT) models to analyze the knowledge states of learners. Participants completed ten adaptive learning sessions, and pre- and post-assessments were compared using paired t-tests to evaluate changes in learning achievement. Results: Among the 50 participants, posttest scores increased compared to pretest scores after 10 days of AI-based tutoring, although the difference was not significant (p > 0.05). A significant correlation was observed between the system scores and the number of solved questions (p = 0.001). In the lower 50% subgroup (n = 25), the posttest scores significantly improved compared to the pre-test scores (p < 0.01). Conclusions: Although the overall score improvements were not statistically significant, the AI tutoring system showed modest potential as a supplementary tool that supports repeated practice and immediate feedback, particularly for students with lower baseline performance. These findings are preliminary and should be interpreted with caution, highlighting the need for further research onenhanced system features and more rigorous study designs.