Novel Algorithm Boosts Stroke Triage, Outcomes Across the US

HELSINKI — Use of a novel algorithm significantly improved prehospital triage for stroke, more accurately directing appropriate patients to centers equipped for endovascular therapy (EVT) and boosting the likelihood of good functional outcomes, based on computer modeling across the continental United States.

“In the MAP-STROKE study, we showed that a novel Bayesian predictive triage algorithm could improve stroke-related neurologic recovery. These improvements were primarily driven by quicker time to endovascular therapy for patients with large vessel occlusion [LVO] strokes, and the magnitude of improvement was most pronounced in rural areas,” lead investigator Santiago Ortega-Gutierrez, MD, clinical professor in the Departments of Neurology, Neurosurgery, and Radiology at the University of Iowa, Iowa City, Iowa, told Medscape Medical News.

“We believe that this new algorithm will get patients to the right hospital faster, matching their specific clinical needs and geography, and if implemented widely, it could lead to around 10,000 additional stroke patients each year achieving a good neurological recovery in the US,” he added.

Santiago Ortega-Gutierrez, MD, presented the findings on May 23 at the European Stroke Organization Conference (ESOC) 2025.

Time-Sensitive Treatment

Stroke is the fifth leading cause of death in the United States, and LVO accounts for over 60% of stroke-related disability, he noted. Outcomes can be significantly improved with EVT, but the treatment is highly time-sensitive — and only about 30% of the US population lives within 30 miles of a center that offers it.

“One of the most important decisions for the half a million Americans transported by ambulance for stroke each year is which hospital they are taken to,” Ortega-Gutierrez said.

Currently, the decision on which hospital a patient with a stroke should be taken to often follows American Heart Association (AHA) guidelines. If an LVO is suspected, the patient should be transported to a center capable of EVT — provided it is within 30 minutes (or 45 minutes in rural areas). Otherwise, the patient is taken to the nearest hospital and may require transfer to an endovascular center if needed, said Ortega-Gutierrez.

“Frequently that complex decision isn’t made by a neurologist or a doctor of any kind; it’s made by a paramedic — many of whom do not have much training — in an ambulance with a lot of diagnostic uncertainty. But that decision, in many cases, sets a patient on a treatment course with far-reaching consequences,” he said.

With funding from the US National Institutes of Health and the National Institute of Neurological Disorders and Stroke, Ortega-Gutierrez and colleagues have spent the past 4 years developing a new algorithm that draws on a range of patient variables available to paramedics — such as age, gender, presenting blood pressure, stroke severity, current medications, and geographic location.

The algorithm estimates the likely stroke type, recommends the most appropriate therapy, and identifies the optimal hospital using GPS data and web-based mapping tools.

Ortega-Gutierrez said the team developed an algorithm that uses readily available patient information to help paramedics identify the likely stroke subtype and determine the most appropriate hospital for treatment.

The current modeling study tested the clinical impact of this new algorithm in a simulated US population and compared outcomes with strategies using the current AHA guidelines for routing decisions or just going to the nearest hospital.

115 Million Stroke Case Scenarios

Drawing on data from previous acute stroke studies, including the Virtual International Stroke Trials Archive and the RACECAT and FAST-MAG prehospital stroke trials, the researchers ran simulations of over 115 million stroke scenarios.

The investigators found that across all stroke events, 63% of cases were triage-mode concordant between the new algorithm and the current guideline-recommended routing.

In many cases, however, the new algorithm recommended a different hospital destination — often involving a slightly longer trip to an alternative facility that was geographically closer to an EVT-capable center.

The study’s primary endpoint was the percentage of patients achieving a good functional outcome as defined by a modified Rankin scale score of 0-2 at 90 days.

Results showed that good functional outcomes were achieved in 56.5% of patients taken to the nearest hospital, 57.1% of those directed by AHA triage guidelines, and 58.6% of those routed using the new algorithm.

When extrapolated to the entire US stroke population, the researchers estimated that the algorithm could lead to 12,000 additional good outcomes per year compared with the nearest hospital strategy and 8500 more than the AHA guidelines.

Patients experiencing an LVO would derive the largest benefit from the new algorithm, with a 4.4% difference in the numbers achieving a good functional outcome compared with the AHA guidelines, Ortega-Gutierrez reported.

The differences in triage outcomes were primarily attributable to shorter time to EVT in patients with LVO stroke, which was reduced by an average of 90 minutes relative to guideline-adherent routing and by 136 minutes relative to nearest hospital transport.

Strong Benefit for LVO Stroke

Results also showed that the new algorithm provided especially strong benefits for patients with LVO stroke in rural areas, reducing time to EVT by an average of 166 minutes compared with guideline-based transport and by 181 minutes compared with going to the nearest facility. This led to a 7.8% increase in the number of rural patients achieving a good neurologic outcome.

The neurologic benefit observed in patients with LVO stroke was offset slightly by a small negative effect for patients without an LVO stroke.

Specifically, patients with ischemic stroke who did not have an LVO had a delay of 14 minutes reaching the hospital and a 4.6% lower rate of getting thrombolysis, translating into a 0.12% reduced chance of achieving a good functional outcome.

For patients with hemorrhagic stroke, there was a delay of 13 minutes in reaching the hospital and a 0.28% decrease in the chance of achieving a good functional outcome.

“So, there’s a very small negative effect on the patients without a large vessel occlusion stroke. But because most of the disability burden in stroke occurs in patients with a large vessel occlusion, there is still an overall benefit on the entire population,” Ortega-Gutierrez said.

“The large benefit for large vessel occlusion patients vastly outweighs the slight disadvantage to the non–large vessel occlusion patients. By using this algorithm, we believe we will be improving the outcomes of the sickest with minimal repercussions on the less sick,” he added.

An obvious limitation of the study is that it is based on computer modeling and efficacy is simulated, which could potentially overestimate real-world performance.

However, the researchers are now planning a real-world clinical trial to try and confirm these results.

Experts Weigh in

Jesse Dawson, MD, professor of stroke medicine at Queen Elizabeth University Hospital in Glasgow, England, emphasized the importance of the MAP-STROKE study findings, noting their relevance as stroke care grows increasingly complex and patients are frequently transferred between institutions.

“This was a modeling study, but it suggested that a relatively small effect can really have quite a transformative impact on patient outcomes at the population level. It will be interesting to see how that can be used in other countries and other healthcare systems,” said Dawson, who was not involved in the study.

Also commenting, Guillaume Turc, MD, professor of neurology at Université de Paris and Sainte-Anne Hospital, Paris, France, described the trial as innovative and a promising step toward improving stroke care but cautioned that the findings rely on certain assumptions — particularly the relatively high estimated proportion of LVO cases — and emphasized the need for confirmation.

The MAP-STROKE study was funded by the National Institute of Neurological Disorders and Stroke. Ortega-Gutierrez was additionally supported by funding from the National Institutes of Health, Patient-Centered Outcomes Research Institute, Stryker, Medtronic, and Methinks. He also received consultant fees from Medtronic and Stryker.