Digital twins and machine learning in stroke monitoring and forecasting

Our vision is to tackle an important societal wellbeing challenge, by creating personalised digital twins for stroke forecasting, and machine learning solutions to analyse stroke imaging data. The work will build on our previous success in stroke modelling. Stroke has a high human cost due to death and disability, and a global economic cost estimated at >$721bn annually. Strokes occur when the blood supply to sections of the brain is interrupted. 85% of UK strokes are ischemic, where a clot, plaque or air bubble blocks a cerebral artery. The remaining 15% of strokes occur due to haemorrhage. The size and location of the blocked or damaged vessel dictates the damage caused by the stroke. Fast stroke identification is important, since rapid treatment improves patient outcomes.

Tools to predict or rapidly identify the onset of a stroke could lead to faster treatment. So, there is a need for technologies to analyse big data arising from continuous monitoring of patients with high stroke risk in the community.  After a stroke occurs, identification of its cause is important, as treatments may be available to reduce risk of additional strokes. Thus, we want to create systems to understand the origins of strokes to improve patient outcomes and predict future stroke risks.

Our ambition is to combine machine learning and simulation technologies, in order to develop personalised digital twins for real-time simulation and analysis of monitoring data and for post-stroke diagnosis from medical imaging. We aim to create healthcare impacts by personalising technologies developed at the OU in collaboration with world-class clinicians and technology companies.