Skip to main content
. 2025 Dec 5;71(11):e20250546. doi: 10.1590/1806-9282.20250546

Table 2. First diagnostic test accuracy.

Participants Responses pa
Correct (%) Incorrect (%)
Human expert 88 12 0.208
ChatGPT-4 80 20
ChatGPT-4o 87 13
ChatGPT o3-mini 89 11
a

Cochran-Q test; no statistical difference was found between paired measurements.