Safe opioid prescribing: a prognostic machine learning approach to predicting 30-day risk after an opioid dispensation in Alberta, Canada.
Opioids are high-risk medications and a significant source of patient harm. Using administrative data for over 390,000 adult patients in Alberta, Canada, who received an opioid prescription from 2017-2018, the authors developed machine learning models to estimate the 30-day risk of opioid-related adverse outcomes. Findings suggest that incorporating hospitalization or physician claims into the models can improve predictive performance, as compared to the inclusion of guidelines or prescribing history alone.