AI Is Saving Lives in Stroke Treatment

According to He, when someone in a remote part of Alaska—about 97% of the state qualifies as remote—suffers a stroke, care providers send CT images to the nearest hospital with stroke diagnosis and treatment capabilities. That process of sending roughly 1,500 images takes about 45 minutes. Specialists then review the images to determine whether the patient needs to be transferred to the hospital for treatment, which usually involves the administering of tissue plasminogen activator (tPA).

In Alaska, that transfer may involve an ambulance, helicopter, and/or fixed-wing aircraft and take hours. Flight crews have 30 minutes to accept the transfer and another 30 to file a flight plan and receive clearance to fly. And after the patient is transported to the hospital, another round of CT images is taken to make sure the patient is still a good candidate for treatment.

Throughout this lengthy process, the patient’s health is declining, reducing the chances that tPA can halt the effects of the stroke and preserve brain function. In some cases, a patient who initially could be saved with tPA might not be saved by the time he or she reaches the hospital.

AI can improve that process, He says, by enabling CT scans to be sent in real time and helping specialists review the images.

Through the RapidAI platform, she says, care providers and specialists can collaborate and share data more quickly and effectively, analyzing a patient’s chances of recovery and giving everyone – care teams, specialists, transport teams and the patient’s family – a more accurate time frame.

“There’s nothing worse than transferring a patient … and then they get here and it’s like, no, they’re not going to be a candidate,” He says. “Now this patient is far from their family [and] their family still has to fly on their own money down to Anchorage. So it’s really about identifying the right patients in a timely manner to make a decision whether it’s transfer or stay.”

From hours down to minutes

According to Jeremy Hunter, CMO and CMIO of the Alaska Native Tribal Health Consortium (ANTHC), one of the participating healthcare networks in the Alaska Stroke Coalition, the coalition sees roughly 150 stroke activations a year. Since joining the project, the amount of time needed to assess and begin treatment has dropped from about four hours to roughly 45 minutes.

Those are telling numbers for a population that can be hundreds of miles from the nearest hospital.

“Without a road system, without reliable connectivity in some places, some without running water in villages, it’s fascinating delivering care up here,” he says.

Hunter says the AI platform gives local providers more confidence in assessing patients. He can use an app to more quickly share data with specialists. An AI interpretation of a scan, he says, can give emergency care providers some vital information on the severity of a stroke within minutes.

Using innovative technology to improve stroke assessment and care isn’t exactly new. Health systems and hospitals across the country, from Chicago to Mississippi, have been using telemedicine and digital health tools for years to improve the process, establishing telestroke networks that connect rural care teams with specialists, even using specially equipped EMS vehicles in large cities to improve emergency diagnosis and care.

But while those advances get patients in front of specialists more quickly, AI tools are helping providers see the data they need to see to make critical decisions.

He says AI can drastically reduce the maddening gaps that affect stroke care, improving the chances that a patient will survive and reducing brain damage caused by those delays. He says AI can help providers understand how much of a patient’s brain has been affected by stroke and what can be saved through intervention. This includes a better understanding of whether a patient can be saved by intervention—a literal pain point when a provider has to decide whether to set up an expensive and stressful emergency transfer for a patient in the throes of a stroke.

An ongoing path to better care

But the technology also gives providers more data, enabling them to understand what causes a stroke and how different treatments work. This can fuel stroke prevention education and resources as well as fine-tuning stroke treatment protocols.

“Really, prevention ultimately is what needs to happen,” she says.

And then there’s cost. Healthcare organizations have little resources to spare on new tech, hence the formation of the coalition and the three-year grant to keep it going. Both Hunter and He say there’s an ongoing effort to sustain this partnership.

“We’ll have financial conversations, but I think it is such a vital tool for improving stroke care that unless it’s an astronomical number that we just simply can’t afford, I don’t see how we can go back to not having it,” Hunter says.

He agrees, saying the ROI for this technology should be measured not only in lives saved and emergency transport and ER costs justified, but in education and other resources that help people reduce their stroke risk and providers understand preventive care, diagnosis and treatment.

“All of that factors into the cost,” she says.