AI-generated Feedback Following Social Robotic Virtual Patient Interactions and Medical Student Performance:...

Area of Science:

• Medical Education Technology
• Artificial Intelligence in Healthcare
• Clinical Reasoning Development

Background:

• Traditional virtual patient (VP) platforms lack individualized feedback, hindering clinical reasoning skill development.
• Large language models (LLMs) offer potential for automated analysis of student-VP interactions and scalable feedback.
• AI-enhanced social robotic VP platforms are promising but require validation for improving clinical performance.

Purpose of the Study:

• To evaluate if AI-generated feedback integrated with social robotic VP interactions enhances medical students' clinical performance.
• To assess improvements in medical history taking and communication skills.

Main Methods:

• A quasi-experimental study involved 115 medical students using a Social AI-Enhanced Robotic Interface for 9 VP cases.
• The intervention group received AI-generated feedback post-interaction, while the control group did not.
• Clinical performance was assessed via an objective structured clinical examination (OSCE) with blinded evaluation.

Main Results:

• Students receiving AI feedback achieved significantly higher OSCE scores (mean difference 0.70, P<.001).
• Significant improvements were noted in generic medical history taking (P=.004).
• The feedback group demonstrated higher pass rates (96.7% vs 79.6%, P=.006).

Conclusions:

• AI-generated feedback integrated into robotic VP interactions significantly boosts medical students' OSCE performance.
• The findings support AI feedback systems as a valuable supplement to expert-led teaching in VP simulations.
• Targeted feedback design is crucial for optimizing competency-specific improvements in medical training.