Wednesday, December 2 | 9:30 a.m.-10:30 a.m. | W3-SSNR10-3 | Room E350
In this presentation, researchers will share how implementing AI software at a comprehensive stroke center didn’t measurably improve patient outcomes.
AI tools have been widely adopted for time-sensitive conditions like stroke, despite a lack of evidence that they improve outcomes, noted presenter Maria Sanmartin, PhD, of the Donald and Barbara Zucker School of Medicine at Hofstra/Northwell. To investigate, the research team compared stroke outcomes before and after five AI tools were introduced at a comprehensive stroke center in the New York metropolitan area.
The researchers retrospectively analyzed 876 patients with acute ischemic stroke treated between 2014 and 2024, excluding those seen while the tools were being rolled out. They used propensity score matching and interrupted time-series analysis to account for differences between the two groups.
After controlling for other factors, the researchers found no significant change in the odds of a good functional outcome at discharge, or in discharge disposition, after AI implementation. The odds of completing CT angiography within 60 minutes of arrival did, however, rise sharply when the AI period began.
“Given prior research demonstrating that AI tools expedite stroke diagnosis, streamline workflows, and enhance collaboration among medical teams, further investigation is warranted to determine whether these changes impact outcomes,” the researchers wrote.
Want to learn more? Get the complete picture at this Monday morning talk.
