Student Perceptions of a Custom Artificial Intelligence Clinical Case Companion

Area of Science:

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

Background:

• Artificial intelligence (AI) tools show potential for enhancing physician assistant (PA) education, particularly in clinical reasoning.
• Limited research exists on custom Generative Pretrained Transformer (GPT) applications within PA curricula.
• This study investigated student experiences with a novel GPT-based clinical reasoning tool.

Purpose of the Study:

• To evaluate first-year physician assistant (PA) students' experiences and perceptions of a custom GPT-4-Turbo clinical case companion.
• To assess the tool's impact on developing clinical reasoning skills across various domains.
• To identify student-reported benefits and areas for improvement in AI-assisted PA education.

Main Methods:

• A mixed-methods study involving 72 first-year PA students utilizing an interactive GPT-4-Turbo clinical case companion.
• Students engaged with Socratic dialogue-based cases in hematology and nephrology.
• Post-engagement surveys (Likert scale, open-ended questions) and thematic analysis were employed.

Main Results:

• An 81.9% response rate (59 students) reported the tool as moderately to extremely helpful (90%).
• Significant improvements were noted in developing differential diagnoses (79.7%), ordering diagnostic studies (81.4%), and interpreting tests (78.0%).
• Key themes included appreciation for immediate feedback (36%) and detailed explanations (36%), with concerns regarding case content accuracy.

Conclusions:

• Custom GPT-based tools can effectively supplement traditional PA education, offering personalized, on-demand learning.
• Students perceived significant gains in clinical reasoning but noted limitations in history-taking skills.
• Future implementation requires faculty oversight and AI literacy training to address accuracy concerns and optimize educational value.