How Data Modelling and AI Are Changing International Stroke Care

Every minute after a stroke matters. Swift treatment to remove or destroy blood clots can dramatically improve outcomes. PenARC brought together researchers, people with lived experience, clinicians and policymakers to substantially improve stroke services and patient lives.

The issue 

Thanks to treatments like clot-busting drugs (thrombolysis) or the mechanical removal of clots (called thrombectomy), patient risk of disability and death following a stroke can be significantly reduced. However, these treatments only work if they’re used quickly and correctly, and there is currently significant variation in how different units use these treatments.   

That’s where the Stroke Audit Machine Learning (SAMueL) project came in. 

What did we do? 

We brought together people with lived experience, NHS clinicians, researchers, and policymakers with the goal of substantially improving services and patient outcomes.  

We used Artificial Intelligence (AI) to look at how different hospitals across the country were using thrombolysis. The results were eye-opening: even when patients had similar conditions, outcomes were better in hospitals that used this treatment more often. 

We didn’t just highlight this discrepancy. We worked with stroke units to enable them to increase appropriate treatment rates, which would result in: 

  • 400 more people each year avoiding lasting disability 
  • 280 fewer people dying or left with severe disability after a stroke each year 

Impact 

This work helped change national policy. NHS England funded work to successfully improve stroke treatment in six underperforming hospitals, increasing thrombolysis rates from 9.7% to 14.5%. This programme has now been expanded to 12 more Trusts. 

And we made sure hospitals can keep improving. Our work was included in the National Clinical Guidelines for Stroke, and the National Stroke Audit incorporates our computer code, helping each unit spot what’s working and what’s not. 

However, thrombolysis is not a suitable treatment for all patients. Thrombectomy – a mechanical procedure to remove clots – must happen in specialist centres. But deciding where these centres should be located is complicated.  

Combining time-to-treatment effects on death and disability with geographical data, we built a model with national policymakers to find the best locations. Conservative estimates suggest that placing these centres according to the model could save 300 people a year from death or disability. 

We’re proud to see our recommendations in action: new centres have been announced in Kent and Norfolk so far. 

David Hargroves, National Clinical Director for Stroke, called PenARC’s work “pivotal in the planning for the delivery of thrombectomy” and said: 

“We are now able to deliver 4,500 thrombectomy procedures per year, which means 1,000 patients are less disabled. PenARC’s work has been critical to our ability to deliver this to the NHS.” 

Learn more about SAMueL 2 – Stroke Audit Machine Learning.

Updated June 2025.