Heston and Lewis conducted a study to evaluate the performance of ChatGPT-4 in risk-stratifying patients with atraumatic chest pain. The researchers compared ChatGPT-4's risk scores with established tools like TIMI and HEART scores using simulated patient data. Although the mean scores correlated well, ChatGPT-4 provided different risk scores for identical patient data when presented on separate occasions. This inconsistency suggests that further refinement and customization are necessary before integrating ChatGPT-4 into clinical practice for cardiac risk assessment.
Citation: Heston TF, Lewis LM (2024) ChatGPT provides inconsistent risk-stratification of patients with atraumatic chest pain. PLOS ONE 19(4): e0301854. https://doi.org/10.1371/journal.pone.0301854