A Novel Big Data Analytics Framework to Predict the Risk of Opioid Use Disorder.

Addiction and overdose related to prescription opioids have reached an epidemic level in the U.S., creating an unprecedented national crisis. This has been exacerbated partly due to the lack of tools for physicians to help predict whether or not a patient will develop opioid use disorder. Prior research lacks the investigation of how machine learning can be applied to a big-data platform to ensure an informed and judicious prescribing of opioids. In this study, we explore the Massachusetts All Payer Claim Data (MA APCD), a de-identified healthcare claim dataset, and propose a novel framework to examine how na\"ive users develop opioid use disorder. We perform several feature engineering techniques to identify the influential demographic and clinical features associated with opioid use disorder from a class imbalanced analytic sample. We then use and compare the predictive power of four well-known machine learning algorithms: logistic regression, random forest, decision tree, and gradient boosting, to predict the risk of such dependency. Results showed that the random forest model outperforms the other three algorithms while determining the features, some of which are consistent with prior clinical findings. We anticipate that this research has the potential for healthcare practitioners to improve the current prescribing practice of opioids, thereby curbing the increasing rate of opioid addiction.