Advancing Medical Artificial Intelligence with Foundation Models (360G-Wellcome-318987_Z_24_Z)

This project aims to drive significant advancements in medical Artificial Intelligence (AI) by developing versatile and efficient medical foundation models. Aligned with the strategic goals of the UK National AI Taskforce and the NHS Long Term Plan, it positions the UK as a leader in both foundation model research and medical AI. The project introduces generalisable technical innovations to optimise raw data distribution and enhance inefficient training strategies. By integrating cutting-edge methodologies with routinely collected multi-modal data from the National Health Service (NHS), the proposed foundation model enhances disease diagnosis and prediction, while unlocking new healthcare applications not previously driven by foundation models. These include treatment recommendations, early marker discovery, and disease subtype identification. The project uses ophthalmology as an exemplar, exemplifying the potential of the foundation model in precision medicine and translation efforts, benefiting both eye health and the emerging field of "oculomics," which uses retinal imaging to gain insights into systemic health. The programme will leverage data from multiple medical sites, including the INSIGHT Health Data Research Hub at Moorfields Eye Hospital, the world's largest ophthalmic imaging bioresource, alongside global research institutes representative of diverse populations, data infrastructure, and disease phenotypes.