Assessing ChatGPT's Capability as a New Age Standardized Patient: Qualitative Study

Area of Science:

• Medical Education Technology
• Artificial Intelligence in Healthcare
• Simulation-Based Learning

Background:

• Standardized patients (SPs) are vital in medical training but resource-intensive and limited in access.
• Artificial intelligence (AI), specifically large language models like ChatGPT, offers potential for virtual SPs to overcome these limitations.

Purpose of the Study:

• To assess medical students' perceptions and experiences using ChatGPT as a virtual SP.
• To evaluate ChatGPT's effectiveness as a virtual SP in a medical school setting.

Main Methods:

• Qualitative study involving 9 medical students interacting with ChatGPT as an SP.
• Pre- and post-interaction surveys and thematic analysis of student responses.
• Observation of live role-play interactions with a predetermined prompt.

Main Results:

• Pre-interaction concerns included prompt engineering, learning efficacy, verisimilitude, and AI accuracy.
• Post-interaction, students noted improved prompt engineering but persistent verisimilitude challenges (lack of empathy, nonverbal cues).
• Students perceived ChatGPT as a valuable supplement, enhancing flexibility and diagnostic skills, with maintained learning efficacy.

Conclusions:

• ChatGPT can augment traditional SPs in medical education, providing accessible and flexible practice.
• Limitations in verisimilitude and prompt engineering prevent ChatGPT from fully replacing human SPs.
• Integrating prompt engineering into curricula and advancing AI are recommended for enhanced virtual SP utilization.