PDG35 Using Machine Learning to Classify Patients on Opioid Use

Abstract

High-frequent opioid use tends to increase an individual’s risk of opioid use disorder, overdose, and death. Thus, it is important to predict an individuals’ opioid use frequency to improve opioid prescription utilization patterns. This study applied five machine learning techniques to predict opioid use frequency, including support vector machine, random forest, neural network, gradient boosting, and XGBoost (extreme gradient boosting). Additionally, this study compared the performance of these machine learning models with penalized logistic regression. This retrospective study included individuals receiving at least one opioid prescription from 2016-2018 in the national representative data, Medical Expenditure Panel Survey. The study outcome measured whether an individual lied in the upper 10 percent of the opioid prescription distribution. The predictors were selected based on Gelberg-Andersen’s Behavioral Model of Health Services Utilization. The prediction performance was assessed using the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC) in the test data. Patient characteristics as predictors for high-frequency use of opioids were ranked by the relative importance in prediction in the test data. Random forest achieved the highest value of both AUROC (0.7726) and AUPRC (0.2871), outperforming logistic regression. In the best performing model, age, the number of chronic conditions, public insurance, and self-perceived health status had enormous predicting power in opioid use frequency. This study demonstrates that machine learning techniques can be a promising and powerful technique in predicting health outcomes.