There’s a saying among doctors who care for stroke patients that “time is brain”, meaning every second an artery is blocked can lead to irreversible damage in brain tissue. And while it applies to timing on the patient side, such as how fast a patient seeks medical attention after symptoms, it is also true in the hospital setting.

To gain precious seconds in how quickly a patient receives treatment, Pittsburgh hospitals are now using apps powered with artificial intelligence to identify patients at risk and immediately notify doctors.

“I like to compare a stroke to a forest fire,” said Dr Raul Nogueira, the new director of UPMC’s Stroke Institute, who came to UPMC in February. “It starts small and burns a few trees and over time will burn the whole forest down. By removing mechanically or dissolving the clot, they put the fire out, but the part of the forest that is burned, we cannot bring back.”

In the past, when patients came into the emergency room with stroke symptoms and were sent for a CT scan to check for abnormalities, they would have to wait for a radiologist to be available to read the images. In January, UPMC started using an app called Viz.ai to read the CT scans using artificial intelligence. If the software detects an issue that needs to be looked at immediately, the app will communicate that urgent need via smartphone to doctors and nurses.

Allegheny Health Network has used a different app with similar functions, called RapidAI, since March of 2020, and first installed artificial intelligence analysis of CT scans for possible stroke patients in 2017. The hospital system analysed data from before and after using artificial intelligence, and doctors presented a paper on their findings at a conference in February.

Doctors tracked the time from when patients got their CT scans to when they actually received incisions for a procedure to unblock the artery, both before the introduction of the artificial intelligence and after. The time dropped significantly, from an average of 93 minutes before use of artificial intelligence to 68 minutes after. Ninety days after their stroke, patients treated after the introduction of the artificial intelligence software also had better scores on a scale measuring their degree of independence.

When a patient in the hospital is deemed to be having a stroke, a notification dings on the phones of app users.

“It’s like a breaking news alert, like CNN or Apple News,” said Dr Russell Cerejo, interim stroke director at Allegheny Health Network and an author of the study. “It all gets sent directly to the stroke team, and we know right away that there is a potential stroke that is happening.”

The artificial intelligence does not replace doctors, who will still read each CT scan, said Dr Nogueira, but it can flag things more quickly and possibly even identify abnormalities a doctor might miss. And because the communication goes through smartphones, using HIPAA-compliant technology, Dr Nogueira said he can respond faster than if he had to receive a phone call or log in to a computer.

UPMC started using the app at UPMC Presbyterian, Mercy and Shadyside in January and is rolling it out to more than two dozen smaller hospitals through June both in and out of the UPMC system, including Excela Health, Heritage Valley Beaver and ACMH in Armstrong County. AHN also uses it at many of its partner hospitals, such as St Vincent in Erie. It is often in the smaller, community hospitals where the apps create the biggest time savings, said Dr Cerejo, because they don’t have as much staff on hand to read CT scans.

The Viz.ai app and the artificial intelligence behind it have been in development for about five years, said Dr Nogueira, who served on the company’s medical advisory board as it began developing the technology in 2018. UPMC began the process of using the technology before Dr Nogueira’s arrival there.

And while the app can help in the hospital, Dr Nogueira emphasised that patients themselves should also be aware of the importance of time in treating strokes.

Stroke experts use the acronym “FAST” to flag the most common stroke symptoms. The “F” is for drooping or asymmetry in the face, the “A” for arm weakness or asymmetry, the “S” for speech changes, such as slurring or difficulty stringing words together, and the “T” means it’s time to call 911.

“Even in this day and age, it's not rare that you would have patients that will come to us very late,” Dr. Nogueira said. “Sometimes this happens because they stay home, they say, ‘I don’t know what’s happening, I’ll just see what happens’. Time is so, so important.”

 - TNS