© 2022 90.5 WESA
Play Live Radio
Next Up:
Available On Air Stations
Health, Science & Tech

A.I. Could Help Predict Who Is At Risk For Opioid Overdoses

Jessica Hill

new study is looking to artificial intelligence to address the opioid crisis by identify people who may be at risk for opioid overdose.

Researchers at the University of Pittsburgh are using machine learning algorithms to analyze data from more than half a million Medicare patients who have at least one opioid prescription. They have found that the program is good at predicting which groups are at risk for opioid overdose, and which groups have little-to-no risk.

Associate professor of medicine and director of Pitt’s Center for Pharmaceutical Policy and Prescribing Walid Gellad explained that identifying low-risk patients could help doctors better target treatments for opioid abuse and overdose.

“You can imagine intervention where, instead of focusing on everyone related to opioids, you just look at those who are really low-risk or really high-risk,” he said.

The mathematical models glean important details like a history of overdose or serious health conditions, which could signal high-risk for opioid overdose.

By applying powerful computing techniques to a complex problem like the opioid crisis, researchers can eliminate some of the uncertainties they currently face.

“We can use all this health care data and instead of just guessing who might be high-risk for opioids or using some kind of crude measure, we can use all the data that’s available in these fancy techniques called machine learning and predictive analytics and try and develop models that better tell us who’s really at high-risk,” said Gellad.

While AI might not solve the opioid crisis, Gellad said that it does have promising applications.

“We have all these powerful computing techniques, we have this opioid problem,” he said. “The idea is, can we use one to help address the other?”

According to Gellad, the researchers are working to incorporate more data sources, which will allow them to predict different outcomes caused by opioids.