In the latest research, it has been found that OpenAI’s GPT-4, an advanced artificial intelligence system, can perform as well as, or even outperform, human ophthalmologists in diagnosing and treating patients with glaucoma and retina diseases.
This study, published in JAMA Ophthalmology, suggests that large language models (LLM) based on artificial intelligence, such as GPT-4, which are trained on extensive data including text and images, could significantly support ophthalmologists in decision-making for the diagnosis and management of glaucoma and retina disorders.
The study, led by Andy Huang, an ophthalmology resident at New York Eye and Ear Infirmary of Mount Sinai Hospital, involved comparing the performance 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. A set of 20 questions (10 each for glaucoma and retina) commonly asked by patients, along with 20 de-identified patient cases from eye clinics, were used in the evaluation.
Results indicated that AI matched or surpassed human specialists in both the accuracy and completeness of medical advice and assessments. Notably, AI demonstrated exceptional proficiency in responding to glaucoma questions and case-management advice, while showing balanced outcomes in retina questions, matching humans in accuracy but surpassing them in completeness.
Dr. Louis R. Pasquale, Deputy Chair for Ophthalmology Research, commented on the surprising efficiency of AI in handling both glaucoma and retina patient cases. The findings suggest that AI could serve as a reliable assistant to eye specialists, providing diagnostic support and potentially alleviating their workload, especially in complex cases or areas with high patient volume. This integration of AI into ophthalmic practice could offer patients quicker access to expert advice, facilitating more informed decision-making for their treatment.