Artificial Intelligence in Retinopathy Screening: A Multinational Validation Study

Abstract

Artificial intelligence tools for diabetic retinopathy screening offer scalable solutions to address global workforce shortages. In a multinational validation study across eight countries, an AI-based retinal image classifier achieved 91 percent sensitivity and 87 percent specificity for detecting referable retinopathy. Performance was consistent across diverse ethnic groups and camera platforms. Integration into community screening programs may reduce preventable vision loss, provided that quality assurance frameworks and referral pathways are in place.