Turkish Abstract Abstract Free Full Text English Similar Articles Mail to Author

¹BS, Department of Ophthalmology, Yale School of Medicine, New Haven, CT
²MD, Department of Ophthalmology, Yale School of Medicine, New Haven, CT
³PhD, Section of Biomedical Informatics and Data Science, Yale School of Medicine, New Haven, CT
⁴MD MBA, Department of Ophthalmology, Yale School of Medicine, New Haven, CT DOI : 10.37845/ret.vit.2023.32.38 Large Language Models (LLMs) have emerged as a potentially transformative force within retinal healthcare, promising substantial advancements that might be analogous to the impact of anti-VEGF injections. These models signify a shift in patient-provider dynamics and clinical documentation, offering avenues to expedite patient inquiries as well as automate documentation through integration with Electronic Health Records. LLMs may increase direct patient engagement and reduce physician burnout. Simultaneously, provider-centric models may aid in navigating intricate clinical scenarios and rare diseases by assisting in literature review. However, their integration poses unique challenges, including the integration of Protected Health Information, interpreting imaging and other information modalities besides text, and the persistent challenge of generating accurate and verifiable responses. These models mandate rigorous evaluation before integration into clinical workflow. As with all medical interventions, there will always be a possibility of negative outcomes, therefore the critical consideration revolves around the acceptable risk of LLMs vs the substantial benefits they may offer. Keywords : Artificial Intelligence, Retina, ChatGPT