OpenAI’s GPT-4 has shown promising results in diagnosing and treating patients with glaucoma and retina disease, potentially outperforming human ophthalmologists. Published in JAMA Ophthalmology, the study reveals how AI-based large language models can assist in decision-making support for ophthalmologists dealing with these common eye disorders.
Lead author Andy Huang from New York Eye and Ear Infirmary of Mount Sinai Hospital states, “The performance of GPT-4 in our study was quite eye-opening.” The AI system was able to match or exceed the expertise of seasoned ophthalmic specialists, indicating its potential in the field.
Researchers compared the knowledge of GPT-4 with 12 attending specialists and three senior trainees from the Department of Ophthalmology at the Icahn School of Medicine at Mount Sinai. They used a set of 20 questions from the American Academy of Ophthalmology and 20 de-identified patient cases to test the AI’s accuracy and completeness in medical advice and assessments.
Results revealed that the AI matched or surpassed human specialists in both accuracy and completeness, with superior performance in glaucoma-related questions and case-management advice. Deputy Chair for Ophthalmology Research, Louis R. Pasquale, notes the AI’s proficiency in handling both glaucoma and retina cases, showing balanced outcomes in diagnosing and suggesting treatments compared to human doctors.
Dr. Huang emphasizes the potential benefits for both eye specialists and patients, stating that integrating AI into ophthalmic practice could lead to quicker access to expert advice and more informed decision-making for treatments. The findings suggest that AI could serve as a reliable assistant in providing diagnostic support and easing the workload of ophthalmologists, especially in complex cases or high patient volume areas.
