Abstract
About this talk
Background: About 1.1 billion people live with vision loss. More than 90% have a preventable or treatable cause, and about 90% live in low- and middle-income countries (LMICs). Artificial intelligence (AI) is promoted as a solution to eye care workforce shortages.
Aim: To examine whether ophthalmic AI is aligned with the global burden of avoidable sight loss, and what is needed for it to improve population eye health equitably.
Methods: Critical narrative review. PubMed was searched on 28 October 2026 with structured queries on global burden and service coverage, AI in LMIC settings, bibliometric analyses, real-world deployment, referral uptake, task-sharing, AI for cataract and refractive error, economic evaluation and dataset equity, supplemented by WHO and IAPB sources. Evidence synthesis: Among adults aged 50 years and over, cataract (15.2 million blind) and under-corrected refractive error (86.1 million with moderate-to-severe impairment) dominate the burden, yet ophthalmic AI and its public datasets concentrate on diabetic retinopathy, glaucoma and age-related macular degeneration, largely from Asia, North America and Europe. Median effective cataract surgical coverage is 14.8% in low-income versus 60.5% in high-income countries, and ophthalmologist density 3.7 versus 76.2 per million. Diabetic retinopathy AI performs well in prospective evaluations in India and Thailand, but referral adherence is the bottleneck: in a Rwandan randomised trial, immediate AI feedback raised adherence from 39.6% to 51.5%, leaving nearly half unseen. A non-AI smartphone system with SMS reminders increased referral attendance from 22% to 54% among Kenyan schoolchildren. An equity-first framework is proposed: prioritise by burden, couple detection to treatment capacity, validate locally, report subgroup performance, and measure effective coverage rather than accuracy.
Conclusion: The binding constraint on global eye health is access to treatment, not detection. AI will contribute most where it is embedded in task-shared pathways with guaranteed onward care, and should be evaluated by effective coverage and vision restored.
Keywords: Global eye health; artificial intelligence; health equity; cataract; refractive error; effective coverage.
Speaker
About the speaker
