Large language models in the management of chronic ocular diseases: a scoping review

Large language models, a cutting-edge technology in artificial intelligence, are reshaping the new paradigm of chronic ocular diseases management. In this study, we comprehensively examined the current status and trends in the application of large language models in major blinding chronic ocular diseases such as glaucoma, cataract, and diabetic retinopathy through a systematic scoping review approach. We conducted this review based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extended to characterize the application of large language models in the field of chronic ocular diseases. The study reveals that large language models demonstrate comparable efficacy to experts in disease screening, diagnostic decision-making, personalized precision treatment recommendation, and accessibility of healthcare resources by integrating multimodal clinical data. However, the application of the technology still faces a triple challenge: (1) the limitation of model generalization ability due to the multimodal nature of clinical data; (2) the ethical controversy caused by the insufficient interpretability of algorithms; and (3) the lack of a standardized validation framework. Future directions emphasize the need for specialized model training, multimodal algorithm optimization, the establishment of a multinational multicenter clinical validation platform, and the construction of an ethical framework for dynamic regulation. Large language models are expected to evolve from an assisted decision-making tool to a core component of precision medicine for chronic ocular diseases, and ultimately to achieve an ecosystem of energy-efficient full-cycle management of chronic ocular diseases.