PharosAI – a partnership between Queen Mary University of London, King’s College London and two NHS Trusts, Guy’s and St Thomas, and Barts Health – has received government funding to develop advanced AI models capable of creating ‘digital twins’ for cancer patients.
PharosAI is one of the UK’s most ambitious programmes in cancer AI. Backed by over £40 million in public and private investment, it aims to combine large-scale, multimodal cancer data and advanced AI models to build a secure, federated AI platform for cancer research.
The new funding allocation will support PharosAI’s DESTINY (DistributEd Digital TwinS of Human Tissues for AI-enabled froNtier biologY) project, which will investigate the possibility of applying AI to routinely collected tissue data in order to predict complex molecular profiling that would otherwise require costly and time-consuming tests.
This could help researchers better understand different malignancies, discover new drugs, improve diagnosis, and develop treatments tailored to individual patients. The grant will allow PharosAI to carry out a phase 1 feasibility study of DESTINY.
This will determine whether the much more ambitious phase 2 project is technically and commercially viable, with successful projects able to apply for up to £10 million.
DESTINY aims to develop frontier models and federated learning methods, which allow AI models to learn from data held by different organisations without the data itself being brought together in one central location.
This approach overcomes the limitation of poor data scale and diversity. The longer-term aim is to use these models to create ‘digital twins’ – digital representations of human tissue that capture information about different biological characteristics.
Professor Anita Grigoriadis, Chief Executive Officer of PharosAI and Professor of Molecular and Digital Pathology at King’s College London, said: “DESTINY is about finding ways to make better use of the vast amounts of biomedical data already being generated across the NHS and research institutions.”
“By combining different types of information and enabling AI models to learn securely across organisations, we hope to build more detailed digital representations of human tissue and predict important biological information.
“If we can show this approach works, it could allow researchers to extract far more insight from existing tissue collections while overcoming some of the practical barriers that currently limit how biomedical data can be shared and used,” Professor Grigoriadis said.
Professor Louise Jones, Chief Scientific Officer of PharosAI and Professor of Breast Pathology at Queen Mary University of London, added: “DESTINY builds on PharosAI’s rich data spanning digital pathology, spatial transcriptomics, whole genome sequencing and proteomics to develop advanced AI models that can predict biological information from routine tissue samples.”
“These digital twins of human tissue could enrich and expand the datasets available for research and, ultimately, transform how molecular analysis is carried out in clinical settings.”
PharosAI is currently seeking to collaborate with organisations developing new oncology treatments and biomarkers to explore how AI could support their research and development activities.
Add Comment