Substance Use Disorder

Tuerk House (TH) is a drug addiction treatment facility serving the Baltimore, MD area. We work in collaboration with TH, Professor Edward Bernat, University of Maryland, and Professor Fadia Shaya University of Maryland Baltimore, School of Pharmacy to analyze the electronic medical record data from TH.

We are currently in the process of creating a rich and multifaceted data set centered on the TH patient population by engineering the data available in the TH EMR. This is supplemented by incorporating other data sources including the Chesapeake Regional Information System for our Patients, PolicyMap, and state law enforcement. This data set captures the complex nature of substance-use disorder, with features spanning psychopathological assessments, neighborhood and socioeconomic traits, some biological testing, and other demographic information.

Our primary approach will then be to apply interpretable machine learning methods in a supervised machine learning context to assess which features are important in predicting treatment and other outcomes such as recidivism, opioid or polysubstance usage, and relapse in a data-driven manner. Our ultimate goal is that this methodology will produce concrete hypotheses of how to inform and improve treatment of substance-use disorder in this population.