Team Evaluates Use of Machine Learning to Predict Opioid Overdose Risk

(April 5, 2019) Researchers from the University of Arizona and others, including Dr. Daniel Malone, RPh, PhD, FAMCP and Dr. Kent Kwoh, MD, recently published a study in JAMA NETWORK OPEN evaluating the use of machine-learning to predict opioid overdose risk. Currently used systems may identify patients who are not truly at a high risk. With a goal of increasing the accuracy for predicting risk of overdose, the research team used machine learning methods that were applied to Medicare data. The results suggest that machine-learning algorithms could predict over use and individuals likely to overuse opioids, especially those in low-risk subgroups. These techniques could be implemented by a variety of healthcare organizations and payers to prevent deaths due to opioid abuse while not over-alerting for individuals who are not abusing opioids.

The research team also included lead author WEI-HSUAN LO-CIGANIC, PHD, former UNIVERSITY OF ARIZONA COLLEGE OF PHARMACY faculty now with THE UNIVERSITY OF FLORIDA SCHOOL OF PHARMACY, along with additional investigators from THE UNIVERSITY OF FLORIDA, UNIVERSITY OF PITTSBURGH, UNIVERSITY OF UTAH, CARNEGIE MELLON UNIVERSITY, both the SALT LAKE CITY and PITTSBURGH VETERANS AFFAIRS HEALTHCARE SYSTEMS, and THE UNIVERSITY OF ARIZONA.

View the article