ResearchProjects

This pilot study examined whether audiovisual data—such as speech, facial expressions, and movement—can improve the prediction of new-onset substance use disorder (SUD) among recently traumatized emergency department (ED) patients. Using data from an NIH-funded study, it developed and compared machine learning models based on electronic health records (EHR) alone versus EHR plus audiovisual data to assess the added predictive value of this novel information source. Findings will inform the development of scalable, ED-integrated tools for early identification and prevention of SUD following trauma.