The first 60 minutes after a stroke are known in medicine as the "golden hour" - and an AI tool now reading brain scans across England has bought patients back more than that.

  • Door-in-to-treatment time cut from 140 to 79 minutes
  • Patients recovering with little or no disability tripled, from 16% to 48%
  • More than 111,000 suspected stroke patients screened so far
  • Deployed across 11 stroke networks in England

How the AI reads a scan

The tool, called Brainomix e-Stroke, was developed in the UK and layered on top of the brain scans already taken when a stroke patient arrives at hospital. Instead of waiting for a specialist to interpret images manually, the AI flags large vessel occlusions - blocked arteries that need urgent mechanical clot removal - within minutes, and lets remote stroke specialists securely review the scan from anywhere in the network.

Government funding through the AI in Health and Care Award, backed by £123 million, has supported the technology's rollout across 11 stroke networks in England, and has helped detect more than 4,500 large vessel occlusions along the way.

What faster diagnosis means for patients

Carol Wilson, a teaching assistant and grandmother, lost her sight and the use of her limbs one morning in June 2021. Doctors used the e-Stroke tool to rapidly identify a blood clot on her brain and recommend an emergency thrombectomy.

"This technology is just amazing. I was able to sit up and text my family later that day, and was back home and able to walk around two days after having a stroke." - Carol Wilson, stroke patient

Health and Social Care Secretary Steve Barclay said AI "has the potential to transform our NHS - delivering faster, more accurate diagnoses and making sure patients can get the treatment they need, when they need it."

Part of a bigger shift in medicine

Stroke care is only one front where AI is speeding up diagnosis in the NHS. At University College London, surgeons used a real-time AI system to analyze video from an operating microscope during a brain tumor removal, a procedure UCL called a world first that helped the surgical team identify risky areas to avoid while removing as much tumor as safely possible.

A 2025 study in JAMA Neurology found a separate AI system, MELD Graph, detected 64% of epilepsy-causing brain lesions that radiologists had missed on first read. NHS officials are careful to describe these tools as decision support rather than a replacement for specialist judgment - but for a condition where, as NHS England's Dr Timothy Ferris put it, "every minute saved during the initial hospital assessment ... can dramatically improve a patient's chance of leaving hospital in good health," shaving an hour off diagnosis is far from a small thing.

Strokes affect roughly 85,000 people in England every year, and the NHS says its AI in Health and Care Award program is continuing to expand the technology to more hospitals nationwide.

Brainomix's own clinical lead, VP of Healthcare Riaz Rahman, says the platform has "fast become a cornerstone" of how integrated stroke networks make treatment decisions, pointing to regions where post-operative functional scores have tripled and access to thrombectomy has more than doubled since rollout. For a diagnosis where speed is everything, that combination of faster scans and faster decisions is what is translating into patients like Carol Wilson walking out of hospital days, not months, after a stroke.