Quest to build the first medical AI foundation model with globally representative data

A research consortium of over 100 study groups in more than 65 countries has launched the Global RETFound initiative, a collaborative effort to develop the first globally representative Artificial Intelligence (AI) foundation model in medicine, using 100 million eye images.

As described in Nature Medicine, the initiative is one of the largest medical AI collaborations ever undertaken, producing one of the most geographically and ethnically diverse medical datasets assembled for AI training purposes. The data will span Africa, the Middle East, North and South America, the breadth of Asia, Oceania, Europe and the Caucasus region.

Led by researchers from Moorfields Eye Hospital NHS Foundation Trust, University College London (UCL), the National University of Singapore Yong Loo Lin School of Medicine (NUS Medicine), and the Chinese University of Hong Kong (CUHK), the consortium will develop an AI model using an unprecedented dataset of over 100 million colour fundus photographs of the retina at the back of the eye, sourced from more than 65 countries.

The initiative builds on the success of RETFound, the first foundation model for retinal and systemic disease detection. Published in Nature in 2023, RETFound was developed by researchers at Moorfields Eye Hospital and UCL Institute of Ophthalmology in London, using 1.6 million retinal images curated by the INSIGHT Health Data Research Hub at Moorfields.

While RETFound has already demonstrated significant potential for medical AI applications, the global model will expand the training data to encompass every continent except Antarctica.

"Current foundational models are trained on data that is geographically and demographically ‘narrow’, which limits their effectiveness and can perpetuate existing health inequalities," explained Dr. Yih Chung Tham, Assistant Professor at NUS Medicine, one of the project leads. "The Global RETFound Consortium addresses this challenge through innovative approaches that enable broad international participation while maintaining strict privacy protections."

A key innovation of the project is its flexible, two-pronged data sharing framework, designed to accommodate varying technical capacities and regulatory requirements across participating institutions. The first approach involves local fine-tuning of generative AI models at individual institutions, with only model weights shared centrally — ensuring no patient data leaves the originating site. The second pathway enables direct sharing of de-identified data through secure infrastructure for institutions that do not have local GPU resources or technical expertise.

Global RETFound will take a two-pronged approach to assembling and curating data, across five stages of foundation model development.

Professor Pearse Keane and his research lab at UCL Institute of Ophthalmology and Moorfields Eye Hospital developed the first AI foundation model in ophthalmology, RETFound, in 2023.

"This dual approach allows participation from research groups regardless of their resource levels," noted Pearse Keane, Professor of Artificial Medical Intelligence at UCL and consultant ophthalmologist at Moorfields. "By combining real and synthetic data generation techniques, we can build a diverse, globally representative dataset without compromising security."

The Global RETFound model will undergo comprehensive evaluation across multiple ophthalmic and systemic diseases, including diabetic retinopathy, glaucoma, age-related macular degeneration and cardiovascular disease. The model will be released under a Creative Commons license, making it freely available for non-commercial research worldwide.

Professor Carol Cheung from The Chinese University of Hong Kong emphasised the broader implications: "This initiative has the potential to establish new international benchmarks for generalisability and fairness in medical AI. By providing researchers worldwide with access to a globally trained foundation model, we can accelerate development of AI tools tailored to local clinical needs with substantially reduced data and computational requirements."

While ophthalmology serves as the initial proof-of-concept, the researchers aim to share their methodologies widely, laying the groundwork for similar global initiatives across other medical specialties.

The project addresses growing concerns about AI bias in healthcare, while demonstrating how international collaboration can advance medical AI development in an equitable way. The consortium welcomes additional researchers and institutions to join their collaborative effort toward more inclusive medical AI development.

The initiative is supported by NIHR Moorfields Biomedical Research Centre and Moorfields Eye Charity.

---