ARVO 2026: Global RETFound study

Global RETFound is an ambitious international collaboration designed to make AI in ophthalmology more equitable, accurate, and globally relevant. In this interview, Paul Nderitu, retinal specialist at Moorfields Eye Hospital and senior research fellow at the UCL Institute of Ophthalmology, explains how the consortium is building a global medical foundation model using retinal images from around the world, including historically underrepresented regions such as Africa, the Middle East, and South America.

To overcome data governance and security barriers, the team uses a dual data-sharing strategy: partners can either contribute synthetic color fundus photographs generated locally from shared code, or send real, anonymized images. This approach has attracted 203 partners across 74 countries, yielding synthetic and real datasets that together power a new foundation model called Global RETFound Beta.

Nderitu highlights that Global RETFound Beta outperforms existing models in detecting ocular and systemic diseases, predicting age from fundus images, and extracting retinal biomarkers. Crucially, the model will be open access for research, enabling teams worldwide—especially in underrepresented regions—to validate performance in their own populations and potentially adopt this framework as a blueprint for future equitable AI models across other medical specialties.