TO THE EDITOR, The recent publication in the American Journal of Emergency Medicine by Günay et al. (2024) presents an intriguing exploration of the presumed diagnostic capabilities of GPT-4 in the domain of electrocardiography assessment. The study ambitiously attempts to juxtapose the proficiency of GPT-4 against that of emergency medicine specialists and cardiologists using multiple-choice questions derived from the “150 ECG Cases” textbook. While the initiative to integrate artificial intelligence into clinical diagnostics is commendable, several aspects of the study design and its presentation raise concerns regarding the interpretability and general applicability of the findings, which we believe merit further discussion. [...]
Niset, A., El Hadwe, S., & Barrit, S. (2024). Did GPT-4 really perform electrocardiography assessment? American Journal of Emergency Medicine, 80, 217-218. https://doi.org/10.1016/j.ajem.2024.04.008 (Original work published 2024)