Progress towards a globally-representative AI model

A research consortium of more than 100 study groups has launched the Global RETFound initiative – a collaborative effort that aims to develop a globally-representative artificial intelligence (AI) model.

Describing the project in Nature Medicine, the researchers describe how the model will use 100 million fundus photographs from 65 countries across every continent except Antarctica.

The consortium will be 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 and the Chinese University of Hong Kong.

Project lead, Dr Yih Chung Tham, of the National University of Singapore, said: “Current foundational models are trained on data that is geographically and demographically ‘narrow’, which limits their effectiveness and can perpetuate existing health inequalities.”

He added: “The Global RETFound Consortium addresses this challenge through innovative approaches that enable broad international participation while maintaining strict privacy protections.”

A statement from Moorfields Eye Hospital highlighted that while the initial-proof-of-concept model will focus on ophthalmology, the consortium aims to share their methodology widely to open up a similar approach to other specialities.

The statement added that the project addresses growing concerns about AI bias within healthcare and aims to advance medical AI development in an equitable way.

Professor Pearse Keane, consultant ophthalmologist at Moorfields Eye Hospital and professor of artificial medical intelligence at UCL, highlighted that the way the data sharing framework is designed enables research groups to participate – regardless of their resource levels.

“By combining real and synthetic data generation techniques, we can build a diverse, globally representative dataset without compromising security,” he said.