ChatGPT-3.5 can effectively recommend surgical treatments for routine glaucoma cases, performing comparably to experienced specialists, according to a study. However, its accuracy decreases in complex cases.
Researchers retrospectively reviewed 60 medical records, categorizing them as “ordinary” (n = 40) or “challenging” (n = 20) cases. ChatGPT was asked to recommend a surgical approach for each case, and its choices were compared with actual patient treatments and the recommendations of three senior glaucoma specialists.
Overall, ChatGPT’s surgical suggestions aligned with patient records in 78% of cases. Agreement with the specialists ranged from 70% to 83%. In ordinary cases, ChatGPT’s performance was statistically comparable to that of experienced surgeons. However, in challenging cases, its accuracy dropped to 65% when compared to patient records, with agreement levels of 50%, 40%, and 70% among the specialists.
The researchers highlight the potential for AI assistance in ophthalmology but stress the need for expert oversight in challenging cases.
Reference
Carlà MM, Gambini G, Giannuzzi F, et al. Testing the Reliability of ChatGPT Assistance for Surgical Choices in Challenging Glaucoma Cases. J Pers Med. 2025;15(3):97. doi: 10.3390/jpm15030097. PMID: 40137413.
Related Content
