AI project aims to reduce disability caused by strokes

A research project from the University of Exeter is using AI and machine learning to personalise stroke care for patients in a bid to reduce future disability.

Stroke is a leading cause of death and disability with more than 100,000 people hospitalised in the UK each year.

The Stroke Audit Machine Learning (SAMuEL) project uses AI to help doctors identify patients most likely to benefit from clot-busting treatment known as thrombolysis, which can reduce disability caused by stroke when given early.

Professor Martin James, consultant stroke physician and honorary clinical professor at the University of Exeter Medical School, said: “SAMueL analysis includes nationwide data from a quarter of a million stroke cases, and by using this data, we can provide each hospital with a tailored target for thrombolysis.

“When teams have used this as a benchmark, they’ve been able to treat more patients, more effectively.

“Stroke has a life-changing impact, so it’s inspiring to see how research like this can lead to more personalised, faster treatment and better outcomes for patients and their families.”

The project is in partnership with the Royal Devon University NHS Foundation Trust and National Institute for Health and Care Research Applied Research Collaboration South West Peninsula (PenARC).

During SAMueL 2, which ran between 2022 and 2024, a tool was developed to better understand how clot-busting drugs are used in hospitals and to help improve their use, so more patients receive the best available treatment as quickly as possible.

The University of Exeter says that this is the world’s first integration of AI into a national stroke audit and it is expected to help thrombolysis treatment be targeted more effectively for local populations.

This treatment has been given to approximately 11% of patients in recent years, including over 1,000 patients per year in the South West.

However, thrombolysis is not suitable for every patient and is only effective if given quickly after a stroke. How often thrombolysis is used, and how quickly it is given, can vary widely across the country.

Building on previous research, the research team at University of Exeter used computer models to study why thrombolysis use varies between hospitals.

They also worked to predict patient outcomes and identify which patient characteristics most influence recovery after a stroke, both with and without thrombolysis.

Methods used in this study to analyse the national stroke audit data may be transferrable to other national clinical audits such as maternity care.

Future studies could explore how machine learning could predict outcomes of thrombolysis for individual people, including the negative effects.

The third stage of research, SAMueL3 began in April 2025 and will run for two years.

It will extend the work to other stroke treatments, including thrombectomy (mechanical removal of a clot) and look at individual brain scans to add more information on who will benefit from, and who might be harmed by, thrombolysis.