The Role of Large Language Models in Ophthalmology: A Review of Current Applications, Performance, and Future...

Large language models (LLMs) are having a significant effect on clinical care as they become increasingly common in society. LLMs are artificial intelligence (AI) systems trained on a large amount of data that use deep learning to understand and generate human language. This enables these models to perform tasks such as answering questions, retrieving necessary information, and summarizing lengthy texts. Ophthalmology represents a unique field where numerous opportunities and innovations can be applied through these models. This review gathers recent studies to assess the current use of LLMs, such as Chat Generative Pretrained Transformer (ChatGPT), in ophthalmic areas, including clinical decision support, documentation, patient education, and medical training. LLMs have shown a remarkable capacity in diagnostic tasks, with GPT-4's performance on ophthalmology board-style assessments and subspecialty diagnostic reasoning matching or almost matching human-level performance. LLMs can also greatly increase workflow efficiency by automating note-taking and patient communication. They can enhance the efficacy and accessibility of medical education by creating educational materials and practice resources. By merging visual inputs with domain-specific data, multimodal and retrieval-augmented models improved their capacity to provide information with greater accuracy and relevance. These technologies are especially helpful in image-rich disciplines, such as ophthalmology, as they can assist in bridging the gap between visual diagnosis and text-based reasoning. Despite its benefits, incorporating AI into clinical care also has serious drawbacks, including the potential for ethical and legal dilemmas, as well as variations in the performance of specialists. LLM trends suggest that advanced, adaptable technologies that maintain anonymity hold considerable promise for enhancing clinical decision-making.