Drug Diversion Detection (Capstone Keynote)

The opioid crisis is rampant in the United States, yet medical professionals are often overlooked as a group that can be vulnerable to opioid addiction. Drug diversion is the act of stealing prescription drugs from a hospital or pharmacy for personal use or sale. It is estimated that up to 1.6% of anesthesiologists will misuse drugs at some point in their career. Drug diversion can lead to patient and self harm, including death, and it can also lead to legal consequences for the healthcare worker. Drug diversion is a serious problem that needs to be addressed.

My team worked closely with subject matter experts to understand the nuances of drug diversion within the context of available data (anonymized surgical data ($n\approx470k$) over the a three-year period from the Duke Health system) to carefully defined our problem. We then developed an two-step ensemble modeling approach to detect drug diversion with high sensitivity. machine learning model to detect potential cases of drug diversion by anesthesiologists from the Duke University Hospitals.

At the end of the project, we presented our method and findings at the Duke MIDS Symposium as the keynote presentation of the year.