A research consortium of more than 65 countries has launched Global RETFound, an initiative to develop the first globally representative artificial intelligence (AI) foundation model in medicine, using 100 million retinal images.
Described in Nature Medicine, the collaboration is led by researchers from Moorfields Eye Hospital, University College London (UCL), the National University of Singapore (NUS) Yong Loo Lin School of Medicine and the Chinese University of Hong Kong (CUHK). The dataset includes colour fundus photographs from Africa, the Middle East, the Americas, Asia, Oceania, Europe and the Caucasus region, making it one of the largest and most diverse medical AI collaborations to date.
The project builds on RETFound, the first AI foundation model for retinal and systemic disease detection, developed by Moorfields and UCL in 2023 using 1.6 million images. Global RETFound extends this to cover multiple ophthalmic and systemic diseases, said a Moorfields statement.
Current foundational models are trained on geographically and demographically narrow datasets, which limits their effectiveness and can perpetuate health inequalities, said Assistant Professor Yih Chung Tham, NUS Medicine. “Global RETFound addresses this challenge through broad international participation while maintaining strict privacy protections.”
Designed to accommodate varying technical capacities and regulatory requirements across participating institutions, Global RETFound’s two-pronged data-sharing framework allows either local fine-tuning of AI models or secure direct sharing of de-identified patient data. “This dual approach allows participation from research groups, regardless of their resource levels,” said Professor Pearse Keane of UCL and Moorfields.
