During a stroke, every moment spent getting to treatment can mean the difference between a good outcome or a devastating one.
A unique, local partnership — powered by machine learning algorithms — holds the key to optimizing the way stroke patients are transported to care facilities.
Every second counts: A typical stroke patient loses 1.9 million neurons each minute a stroke is left untreated. (Image courtesy of Austin-Travis County EMS)
No two strokes are the same, and each carries a unique level of risk and need: While some strokes may be treated quickly with medication, others require emergency surgery and a specialized facility. When paramedics decide where to take a patient, they must weigh the benefit of transporting to a more specialized facility versus the cost of time to get there — and every second matters.
A unique partnership between Austin-Travis County Emergency Medical Services and researchers at Dell Medical School is aiming to ensure that every patient is delivered to the optimal care facility, every time.
“The longer it takes to get a stroke patient to treatment, the more brain matter you lose,” says David Paydarfar, M.D., Ph.D., chair of the Department of Neurology at Dell Med, who leads the work. “What we’re looking to do is aid EMS partners in reducing some of the complexity of second-by-second decision-making as they try to get each patient to the right place to be treated.”
Thanks to artificial intelligence modeling supported by a data brokerage agreement between Dell Med, ATCEMS and Ascension hospitals in the region, the team used local patient data to build models that can predict the optimal delivery point for a person having a stroke anywhere in Travis County, depending on their needs. These models will help inform both transport policies and potential on-the-ground, real-time tools for emergency responders.
“There’s no replacement for our skilled first responders and the expertise they bring to these sensitive, ever-changing situations,” says Mark Escott, M.D., chief medical officer and EMS system medical director for the city of Austin, as well as the program director for Dell Med’s EMS Fellowship. “But partnerships like these, where we have the opportunity to collaborate with world-class researchers and engineers, mean that we can create precision tools and policies that allow our paramedics to focus on the most important job at hand: taking care of the person in front of them.”
Real Data, Real Insights
The data agreement is an essential part of the partnership, allowing for more locally accurate predictions than models based on mock patient data.
“Working with EMS and Ascension, we were able to use real data like 911 calls, ambulance records, dispatch notes and patient information to get detailed insights into what the processes and outcomes were,” says Joshua Chang, M.D., Ph.D., assistant professor in the Department of Neurology. “Ideally what we want to do is provide EMS teams with something like a real-time app that takes into account patient information alongside live data like traffic or hospital volumes, and gives them a fast, verifiable recommendation of where to go.”
Closing Disparities, Here & Beyond
The availability of sites that can perform specialized procedures means that patients in Travis County are usually at low risk for transport to less capable facilities. But in more rural areas, the margin of error is smaller, and tools created using the team’s algorithms could be scaled elsewhere.
“At what point do you drive that further distance to go to the more specialized hospital?” Chang says. “One of the things this data can inform is how to set policies around those thresholds for EMS partners outside of just Austin and Travis County. And we can also help advocate for resources like mobile stroke units or upgrading hospital resources in strategic areas so that transport time isn’t the deciding factor in outcomes for these patients.”