A UK cancer AI consortium wants to know whether algorithms can read complex biology out of tissue samples hospitals already store, and Innovate UK has put money behind the question. PharosAI, a partnership spanning Queen Mary University of London, King’s College London and the NHS trusts Guy’s and St Thomas’ and Barts Health, will pursue the work through a project called DESTINY.
The effort has attracted more than 40 million pounds in public and private backing to date. DESTINY will test whether AI can infer molecular information that today demands costly, slow tests, with the eventual goal of building a digital twin of human tissue: a representation that captures its many biological characteristics.
Capturing tissue properly means weaving together microscope images, DNA sequences, gene activity and protein measurements with patient clinical records. Models trained across those combined sources might expose biology that no single source reveals on its own. To get there, the project will rely on federated learning, an approach that lets models learn from data kept at separate institutions without centralizing it. That method is less proven with mixed biological and medical data than with simpler datasets.
This award bankrolls a feasibility study. Projects that advance from this stage of Innovate UK’s Frontier AI Discovery competition become eligible to seek up to 10 million pounds in further funding.
