Scientists have developed a novel smartphone application that uses artificial intelligence to diagnose a stroke in less than four minutes.
The tool does so by analyzing a patient’s speech patterns and facial movements and can make the determination with the accuracy of an ER doc, researchers claim.
“Brain is time,” as the saying goes. And emergency providers have precious moments to triage, deciding whether to send patients to a neurologist or for an often “expensive and time-consuming radioactivity-based scans,” noted investigator James Wang.
“Currently, physicians have to use their past training and experience to determine at what stage a patient should be sent for a CT scan," Wang, PhD, a professor of information sciences and technology at Penn State, said in a statement issued Oct. 22. "We are trying to simulate or emulate this process by using our machine learning approach."
To develop their app, researchers trained a computer model using data from more than 80 patients experiencing stroke symptoms at Houston Methodist Hospital. They asked each individual to perform a test, examining their speech and cognitive communication while recording via smartphone. Wang and colleagues claim they are the first to analyze the presence of stroke among actual ER patients using natural language processing to pinpoint abnormalities such as slurred speech.
Testing out the model on patients, they found it measured up well against ER docs, including a 93.12% sensitivity rate while maintaining 79.27% accuracy. Experts hope this intervention will help providers strike the right balance between over-using CT scans and underdiagnosing this concern.
“If we can improve diagnostics at the front end, then we can better expose the right patients to the right risks and not miss patients who would potentially benefit,” said John Volpi, MD, a vascular neurologist and co-director of the Eddy Scurlock Stroke Center at Houston Methodist, said in a statement. "We have great therapeutics, medicines and procedures for strokes, but we have very primitive and, frankly, inaccurate diagnostics."
Volpi and colleagues recently presented their paper at the International Conference on Medical Image Computing and Computer Assisted Intervention. You can read more about their work here. Penn State and Houston Methodist are also jointly pursuing a patent for the app.