Time Is Brain

Artificial Intelligence > Time Is Brain

Time Is Brain

Mayo Clinic experts are leveraging AI to transform stroke care and outcomes

By Alison Caldwell, Ph.D. Photography by Paul Flessland Illustrations by Beth Goody

Share

Share this:

In stroke care, time is brain.

From the moment a stroke begins until clinicians have restored normal blood flow in the brain, the clock is ticking. Every minute can mean the death of millions of neurons — our brain’s fragile, critical cells that control our every function. But it takes time to determine if a patient is experiencing a stroke and, if so, what kind of stroke they are having, before any treatment can be given. When it comes to improving stroke care, anything that can speed up the process of identifying and addressing the condition can have an enormous impact on patient recovery.

Enter: Artificial intelligence (AI). At Mayo Clinic, clinicians are using AI algorithms to speed up stroke detection and diagnosis and coordinate care teams to get patients the treatment they need sooner, saving millions of brain cells and improving patient outcomes.

A Ticking Clock

For Sophia Chan, that clock started ticking at approximately 2:05 on a Thursday afternoon in February 2022. It was just another day at her high-pressure job as a television producer when she began to experience a severe headache.

She explains that she doesn’t remember what happened that day — she only knows what she’s heard from her husband and the first responders. “Normally I would just go lie down for a bit and hope for the headache to go away,” Sophia says. “Apparently that day I did end up calling for help, and that’s what saved my life.”

By the time the first responders arrived just minutes later, she was unresponsive. She’d suffered a brain aneurysm while she was home alone. She was transported to Mayo Clinic in Jacksonville, just 15 minutes away.

While Sophia can’t remember anything from that day, her husband, Bobby Cullen, remembers it all in vivid, painful detail, recalling how he spoke to her just hours before a neighbor texted to ask about the ambulances outside their house.

He had been out of town for work, and it took him five hours to get back home. By the time he arrived, Sophia had already been in and out of surgery, and her status was still so tenuous that he wasn’t allowed into her room. He was finally allowed to see her in the earliest hours of the morning, when a clinician told him that it was time to say goodbye — Sophia wasn’t expected to survive for much longer. “I stood by her head for the next eight hours,” Bobby says. “The doctors tried to tell me to sit down, but I told them I wouldn’t sit down until she’d taken her last breath.”

Those eight hours turned into 36 hours, then 72. “On day 7, one of the doctors told me that I should go home and take care of myself,” says Bobby. Sophia continued to beat the odds, but she wasn’t out of the woods. It took almost three weeks for her care team to fully stabilize her condition. She can’t remember anything that happened during the first 21 days of her hospitalization.

Accelerating Diagnosis and Intervention

William Freeman, M.D., is focused on developing systems to better recognize and treat hemorrhagic strokes like Sophia’s, which are less common but often more debilitating than ischemic strokes. This is because ischemic strokes are the result of a blocked blood vessel, which leads to brain tissue injury as the cells are deprived of oxygen. Hemorrhagic strokes, caused by a ruptured blood vessel, result in rapid tissue damage as blood pools and increases pressure in the brain.

The treatments for the two types of strokes are different. Treatment for an ischemic stroke involves breaking up the blood clot, either with a medication or mechanically, so the blood can flow again. Hemorrhagic strokes are treated by providing medications to reverse any blood thinners the patient may be on, stop the bleeding, and relieve pressure on the brain to reduce tissue damage.

Knowing which kind of stroke a person is having is very important for choosing the right treatment. The wrong treatment could make a patient’s condition worse.

With the help of AI, a process that once could take half an hour or longer can now take just seconds. When a patient enters the emergency department with a suspected stroke, the first critical step is getting them a CT scan, generating detailed images of the patient’s brain using X-rays. Technicians then review the images looking for abnormalities in the brain to determine the location of the stroke and what type of stroke it is.

Now, an AI algorithm trained on a database of CT images from patients who have had strokes can rapidly scan hundreds of images and pull out the ones showing an abnormality. A technician reviews the relevant images and confirms the algorithm’s assessment. “A full CT and CT angiogram can be up to about 1,200 pictures,” says Dr. Freeman. “And before the software came into play, we’d be looking manually, slice by slice. It seems like it takes an eternity. AI can compress all those minutes down into seconds.”

After diagnosis comes intervention. Once the stroke has been located and clinicians determine its type, a care team is gathered to initiate the appropriate treatment. Dr. Freeman says AI can smooth this process too by sending automated messages to on-call clinicians as soon as a diagnosis is reached.

“Now, I’ll be in the CT control room, and while a technician is still processing the images, my smartphone pops up with a notification telling me that it’s go time,” he says. “It’s really a sight to behold.”

42 Million Neurons Saved With AI Intervention

30 MINUTES

Patient experiences ischemic stroke symptoms, such as slurred speech, numbness or weakness on one side of the body, or a sudden severe headache.

+20 MINUTES

EMT services are called for the patient, and patient is transported to hospital by EMTs.

+10 MINUTES

Patient arrives at hospital and is triaged in the emergency department. Care team suspects a stroke, and the patient is sent for a CT scan.

+10 MINUTES

Patient undergoes CT scan, and images are sent for review.

WITHOUT AI INTERVENTION

+25 MINUTES

CT images are manually reviewed by a technician to identify the stroke site and type. Care team is gathered to provide treatment. Patient receives treatment for stroke and begins the recovery process.

TOTAL TIME: 95 MINUTES

TOTAL NEURONS LOST: 180.5 MILLION

WITH AI INTERVENTION

+3 MINUTES

CT images are run through AI algorithm to identify stroke site and type. AI system alerts care team and directs them to the patient for treatment. Patient receives treatment for stroke and begins the recovery process.

TOTAL TIME: 73 MINUTES

TOTAL NEURONS LOST: 138.7 MILLION

The Difference a Minute Can Make

While AI tools have been well studied in ischemic stroke, they are less developed for use with hemorrhagic stroke, and that’s what Dr. Freeman wants to change.

“In a hemorrhagic stroke, patients get super sick, super fast,” he explains. “We estimate that patients lose between 6 million and 8 million brain cells per minute in just the first two hours. AI can help get patients out of the waiting room and into treatment much faster.”

Research has found that integrating AI into care for an ischemic stroke can save an average of about 22 minutes. With an estimated 1.9 million neurons lost during every minute of an ischemic stroke, this adds up to about 42 million neurons saved. With even more neurons lost per minute in hemorrhagic stroke, saving even just 10 minutes could have a dramatic impact on a patient’s recovery.

This technology, along with training and teamwork, is already having an impact in the clinic. “With AI implementation, we’re absolutely seeing a difference,” says Kacie Brewer, P.A.-C., who is a member of Sophia’s care team. “It’s getting patients the care they need faster by speeding up the diagnosis and pulling together the right team as quickly as possible.”

Sophia is acutely aware of the importance of those 10 minutes. Her proximity to Mayo Clinic and the AI algorithms that allowed the care team to find her aneurysm in minutes are likely the keys to her remarkable recovery. “I feel very fortunate to have been near Mayo Clinic,” she says. “The care I received was the best of the best.”

Her treatment at Mayo Clinic did more than just save her life. Many who survive hemorrhagic strokes go on to have significant lifelong disabilities. Two years after her stroke, Sophia is thriving, getting back to her active lifestyle of chasing around her sporty 9-year-old twins and easing into her yoga practice.

The family has moved to California to be closer to family, but still travels back to Jacksonville for follow-up visits.

Sophia’s case is remarkable because it’s still not yet the norm — most patients who suffer a hemorrhagic stroke do not experience a recovery like hers. Dr. Freeman believes that better AI algorithms and implementations, along with other cutting-edge technologies, can change that.

Sophia is now participating in the DISCOVERY study, a clinical trial leveraging Mayo Clinic’s expertise in neurology and neuroimaging to understand the risks of post-stroke cognitive impairment in diverse populations. She hopes that the research can lead to better outcomes for other patients like her.

And, every year, she and Bobby make a special effort to send treats and thank each of the doctors and nurses who provided so much of the care and support their family received during those first few impossible weeks.

“We really believe in Mayo Clinic’s values and mission,” says Bobby. “We could see how all of Sophia’s care was really a group effort, and it saved her life.”