Detecting Opioid-Related Aberrant Behavior using Natural Language Processing

The United States is in the midst of a prescription opioid epidemic, with the number of yearly opioid-related overdose deaths increasing almost fourfold since 2000¹. To more effectively prevent unintentional opioid overdoses, the medical profession requires robust surveillance tools that can effectively identify at-risk patients. Drug-related aberrant behaviors observed in the clinical context may be important indicators of patients at risk for or actively abusing opioids. In this paper, we describe a natural language processing (NLP) method for automatic surveillance of aberrant behavior in medical notes relying only on the text of the notes. This allows for a robust and generalizable system that can be used for high volume analysis of electronic medical records for potential predictors of opioid abuse.