Assessing ChatGPT’s Clinical Competency and Patient Perceptions in Emergency Medicine: Insights from Clinical Performance Examinations (Preprint)

Abstract

<sec> <title>BACKGROUND</title> Emergency medicine can benefit from AI due to its unique challenges, such as high patient volume and the need for urgent interventions. However, it remains difficult to assess the applicability of AI systems to real-world emergency medicine practice, which requires not only medical knowledge but also adaptable problem-solving and effective communication skills. </sec> <sec> <title>OBJECTIVE</title> We aimed to evaluate ChatGPT's performance in comparison to human doctors in simulated emergency medicine settings, utilizing the framework of Clinical Performance Examination (CPX). </sec> <sec> <title>METHODS</title> Twenty-eight text-based cases and four image-based cases relevant to emergency medicine were selected. Twelve human doctors were recruited to represent the medical professionals. Both ChatGPT and the human doctors were instructed to manage each case like real clinical settings with simulated patients. After the CPX sessions, the conversation records were evaluated by an emergency medicine professor on history taking, clinical accuracy, and empathy on a 5-point Likert scale. Simulated patients completed a 5-point scale survey including overall comprehensibility, credibility, concern reduction for each case. Additionally, they evaluated whether the doctor they interacted with was similar to a human doctor. The mean scores from ChatGPT were then compared to those of the human doctors. </sec> <sec> <title>RESULTS</title> ChatGPT scored significantly higher than the physicians in both history-taking (mean score 3.91 [SD 0.67] vs. 2.67 [SD 0.78], P &lt; 0.01) and empathy (mean score 4.50 [SD 0.67] vs. 1.75 [SD 0.62], P &lt; 0.01). However, there was no significant difference in clinical accuracy. In the survey conducted with simulated patients, ChatGPT scored higher for concern reduction (mean score 4.33 [SD 0.78] vs. 3.58 [SD 0.90], P = 0.04). For comprehensibility and credibility, ChatGPT showed better performance, but the difference was not significant. In the similarity assessment score, no significant difference was observed (mean score 3.50 [SD 1.78] vs. 3.25 [SD 1.86], P = 0.71). </sec> <sec> <title>CONCLUSIONS</title> ChatGPT’s performance highlights its potential as a valuable adjunct in emergency medicine, demonstrating comparable proficiency in knowledge application, efficiency, and empathetic patient interaction. These results suggest that a collaborative healthcare model, integrating AI with human expertise, could enhance patient care and outcomes. </sec>

Ähnliche Arbeiten