Using machine learning to predict risk of incident opioid use disorder among fee-for-service Medicare beneficiaries: A prognostic study

• Published in: PloS one Vol. 15; no. 7; p. e0235981
• Main Authors: Lo-Ciganic, Wei-Hsuan, Huang, James L., Zhang, Hao H., Weiss, Jeremy C., Kwoh, C. Kent, Donohue, Julie M., Gordon, Adam J., Cochran, Gerald, Malone, Daniel C., Kuza, Courtney C., Gellad, Walid F.
• Format: Journal Article
• Language: English
• Published: San Francisco Public Library of Science 17.07.2020; Public Library of Science (PLoS)
• Subjects: Addictions; Algorithms; Artificial neural networks; Beneficiaries; Biology and Life Sciences; Buprenorphine; Computer and Information Sciences; Demographics; Demography; Diagnosis; Drug addiction; Drug overdose; Drug use; Epidemiology; Government programs; Health aspects; Health care policy; Health risk assessment; Heroin; Internal medicine; Learning algorithms; Machine learning; Medical diagnosis; Medical prognosis; Medicare; Medicine; Medicine and Health Sciences; Methadone; Methods; Narcotics; Neural networks; Opioid abuse; Opioids; Performance prediction; Pharmaceuticals; Pharmacy; Physical Sciences; Predictions; Prescription drugs; Research and Analysis Methods; Risk; Risk factors; Samples; Social Sciences; Statistical methods; Statistics; Subgroups; Substance use disorder; Technology application; Training; United States; Pittsburgh Pennsylvania; Florida; United States > US; Utah; Arizona; Pennsylvania
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0235981

To develop and validate a machine-learning algorithm to improve prediction of incident OUD diagnosis among Medicare beneficiaries with [greater than or equal to]1 opioid prescriptions. This prognostic study included 361,527 fee-for-service Medicare beneficiaries, without cancer, filling [greater than or equal to]1 opioid prescriptions from 2011-2016. We randomly divided beneficiaries into training, testing, and validation samples. We measured 269 potential predictors including socio-demographics, health status, patterns of opioid use, and provider-level and regional-level factors in 3-month periods, starting from three months before initiating opioids until development of OUD, loss of follow-up or end of 2016. The primary outcome was a recorded OUD diagnosis or initiating methadone or buprenorphine for OUD as proxy of incident OUD. We applied elastic net, random forests, gradient boosting machine, and deep neural network to predict OUD in the subsequent three months. We assessed prediction performance using C-statistics and other metrics (e.g., number needed to evaluate to identify an individual with OUD [NNE]). Beneficiaries were stratified into subgroups by risk-score decile. The training (n = 120,474), testing (n = 120,556), and validation (n = 120,497) samples had similar characteristics (age [greater than or equal to]65 years = 81.1%; female = 61.3%; white = 83.5%; with disability eligibility = 25.5%; 1.5% had incident OUD). In the validation sample, the four approaches had similar prediction performances (C-statistic ranged from 0.874 to 0.882); elastic net required the fewest predictors (n = 48). Using the elastic net algorithm, individuals in the top decile of risk (15.8% [n = 19,047] of validation cohort) had a positive predictive value of 0.96%, negative predictive value of 99.7%, and NNE of 104. Nearly 70% of individuals with incident OUD were in the top two deciles (n = 37,078), having highest incident OUD (36 to 301 per 10,000 beneficiaries). Individuals in the bottom eight deciles (n = 83,419) had minimal incident OUD (3 to 28 per 10,000). Machine-learning algorithms improve risk prediction and risk stratification of incident OUD in Medicare beneficiaries.