Machine Learning Based Opioid Overdose Prediction Using Electronic Health Records.

摘要: Opioid addiction in the United States has come to national attention as opioid overdose (OD) related deaths have risen at alarming rates. Combating epidemic becomes a high priority for not only governments but also healthcare providers. This depends on critical knowledge understand risk of patients. In this paper, we present our work building machine learning based prediction models predict patients history patients' electronic health records (EHR). We performed two studies using New York State claims data (SPARCS) with 440,000 and Cerner's Health Facts database 110,000 Our experiments demonstrated that EHR can achieve best recall random forest method (precision: 95.3%, recall: 85.7%, F1 score: 90.3%), precision deep 99.2%, 77.8%, 87.2%). discovered clinical events are among features predictions.