High-volume ophthalmic services face growing pressure from aging populations and chronic eye disease burdens. Traditional care models struggle to meet the demand – globally, over 2.2 billion people have a vision impairment and specialist resources remain limited. Recent advances in large language models (LLMs), offer transformative potential to augment eye care delivery. This presentation will introduce LLMs in healthcare and ophthalmology, illustrating how these intelligent systems can interpret clinical data, converse naturally, and support decision-making. We will survey cutting-edge global applications of LLMs in ophthalmology – from automated retinal image interpretation and screening triage, to clinical decision support and multi-modal data integration combining text and images. Notable benchmarks demonstrate LLM capabilities approaching specialist-level performance. At the same time, we address current challenges including data privacy, ethics, and real-world implementation barriers. The talk aligns these innovations with China's national AI strategy and with Hong Kong's digital health policies for use of generative AI. We will share our institution's experience in developing ChatZOC, China's first ophthalmic LLM, which achieves state-of-the-art performance in ophthalmic tasks, and a successful LLM-powered eye health assistant that has facilitated over a million patient interactions. Finally, we highlight the tangible impacts on care delivery, efficiency, and patient outcomes, and discuss future directions and collaboration opportunities to harness intelligent solutions for accessible, efficient, and high-quality ophthalmic care at scale.
