A Clinical Evaluation of Cardiovascular Emergencies: A Comparison of Responses from ChatGPT, Emergency Physicians, and Cardiologists.

Artificial intelligence (AI) tools, like ChatGPT, are gaining attention for their potential in supporting clinical decisions. This study evaluates the performance of ChatGPT-4o in acute cardiological cases compared to cardiologists and emergency physicians. Twenty acute cardiological scenarios were used to compare the responses of ChatGPT-4o, cardiologists, and emergency physicians in terms of accuracy, completeness, and response time. Statistical analyses included the Kruskal-Wallis H test and post hoc comparisons using the Mann-Whitney U test with Bonferroni correction. ChatGPT-4o and cardiologists both achieved 100% correct response rates, while emergency physicians showed lower accuracy. ChatGPT-4o provided the fastest responses and obtained the highest accuracy and completeness scores. Statistically significant differences were found between ChatGPT-4o and emergency physicians ( < 0.001), and between cardiologists and emergency physicians ( < 0.001). A Cohen's kappa value of 0.92 indicated a high level of inter-rater agreement. ChatGPT-4o outperformed human clinicians in accuracy, completeness, and response time, highlighting its potential as a clinical decision support tool. However, human oversight remains essential to ensure safe AI integration in healthcare settings.