Key points from article :
Researchers from the University of Exeter, led by Professor Martin James, are using artificial intelligence (AI) and machine learning to improve stroke treatment through a project called SAMueL-2, published as part of ongoing work supported by the NIHR Applied Research Collaboration South West Peninsula (PenARC). In partnership with the Royal Devon University NHS Foundation Trust, the team developed a tool that helps doctors decide which patients are most likely to benefit from clot-busting medication (thrombolysis), aiming to reduce long-term disability after stroke.
Thrombolysis is a time-sensitive treatment that dissolves blood clots, but it’s not suitable for every patient and is only effective if administered quickly. Despite its benefits, use of the treatment varies significantly across UK hospitals. SAMueL-2 addresses this by analysing data from over 250,000 stroke cases to identify how hospitals use thrombolysis and offer tailored improvement targets. Hospitals that adopted these data-driven benchmarks have already shown better treatment rates.
The researchers used explainable machine learning to understand why some hospitals treat more patients with thrombolysis than others, and what patient characteristics most affect recovery. The team also developed a web app to help hospitals compare and adjust their thrombolysis decisions and timing. This is the first time AI has been integrated into a national stroke audit, and the team hopes to expand its methods to other areas of healthcare, such as maternity services.
Ultimately, this work aims to personalise and speed up stroke treatment, improving patient outcomes nationwide. It also reflects a broader effort to make AI-driven tools practical and accessible for clinicians on the ground.
