UKRI announces £13 million funding for artificial intelligence in health, with three projects launched at Imperial College London.
Imperial College London researchers have been awarded almost £1.8m as part of a new stream of funding from UK Research and Innovation (UKRI) to explore how artificial intelligence (AI) can improve health research.
The AI for Health funding stream announced today is part of the UKRI's Technology Missions Fund, with Imperial's funded projects to be led by Professor Fan Chung, Dr James Kinross and Professor Daniel Rueckert.
"This funding from UKRI will enable Imperial researchers to explore and exploit the benefits of AI for health, patients and society – spanning areas as varied as cancer, sight loss and air pollution." Professor Mary Ryan Vice-Provost (Research and Enterprise)
The interdisciplinary projects are part of a nationwide package of funding granted to revolutionise AI healthcare research.
Professor Mary Ryan, Vice-Provost (Research and Enterprise) at Imperial College London, said: "Application of artificial intelligence tools will have far-reaching impact on society, and we are seeing growing investment from industry and governments internationally to try to harness its potential. This funding from UKRI will enable Imperial researchers to explore and exploit the benefits of AI for health, patients and society – spanning areas as varied as cancer, sight loss and air pollution."
"Imperial is at the forefront of AI research and through Imperial initiatives such as I-X and the Data Science Institute we are carrying out both the basic research to create safe and trusted AI, as well as applying AI to solve some of the world's most pressing challenges in health, sustainability and security."
Modelling our environment
The AI-Respire project, which won £600,000 funding through the new scheme, aims to develop detailed models of real-world environments to help solve some of the health challenges posed by air pollution.
Led by researchers from the National Heart & Lung Institute, School of Public Health and the Faculty of Engineering, the project will generate complex AI models, to incorporate big data from health cohorts to link exposure to respiratory outcomes and cellular responses to pollution, as well as air quality and weather data.
Data taken from devices such as smart watches and mobile phones, combined with modelling of how air flows in urban areas and how the small particle pollutants spread via air currents, will help to uncover the extent of air pollution's impact at the individual level.
The researchers hope insights gained could ultimately help to improve diagnoses of health issues.
Professor Fan Chung, Head of Experimental Studies Medicine at Imperial's National Heart & Lung Institute, said: "Air pollution represents a major global challenge and a threat to human health. By modelling the complex interaction between pollutants, the built environment, and human health, we hope this AI-assisted approach will enable us to gain deeper insights into the impacts of air pollution on health, and how we can reduce this. It will also help those affected by pollution to build up their resilience.''
Medical guidance
The INDICATE project will see researchers from Imperial's Faculty of Medicine and the Faculty of Engineering will investigate whether artificial intelligence (AI) can improve how we generate medical guidance, by scanning, curating and fact-checking the available medical literature.
Bringing together expertise from the Department of Surgery and Cancer, The Department of Computing, and the Data Science Institute, the team will test if its approach could help to increase the speed and efficiency of clinical guidance creation for breast cancer treatment.
Through this initial work, the researchers hope to validate an AI-enhanced tool capable of trawling and filtering the medical literature, detecting and eliminating fraudulent or flawed research, and generating a real time summary of the available evidence, providing clinicians with the best possible evidence base for making treatment decisions.
The project won almost £570,000 of funding.
Dr James Kinross, Reader in Surgery in the Department of Surgery & Cancer, said: "Thousands of clinical research papers are published each day in any one area, but their relevance and quality can vary greatly. This means crucial evidence reviews to develop clinical guidance can be time and resource intensive. Using an AI-assisted model, our approach has the potential to improve this process significantly, providing clinicians and decision makers with the most up to date summary of the best available evidence, which could ultimately improve treatment and outcomes for patients."
Tracking eye health
In a third project, to be led by Professor Daniel Rueckert from the Department of Computing and backed by £600,000 in funding, researchers will explore how AI could be used to track eye health. The project will develop a privacy-preserving AI tool to improve the diagnosis and assessment of retinal fibrosis – a condition that can affect the effectiveness of treatments for age-related macular degeneration (AMD), a leading cause of vision loss.
While current treatments can have a significant impact on reducing the levels of blindness and restoring vision in AMD, their effectiveness is limited by fibrosis of the retina. Currently, ophthalmologists assess retinal fibrosis and disease activity manually, with a wide variation between clinicians in how they assess the degree of fibrosis.
By using AI, the team hopes to automate and improve how fibrosis is detected and characterised in eye scans, with the hope of standardising its clinical assessment, improving speed and accuracy and leading to better patient outcomes.
Professor Daniel Rueckert, Professor of Visual Information Processing from the Department of Computing, said: "Medical imaging is one field where the innate pattern-recognition strengths of AI can be readily leveraged. Incorporating AI into the assessment of retinal fibrosis has the potential to improve accuracy of classifying what stage disease is at, and how it's likely to progress, which will better inform treatment decisions for the patient."
