Substance Use Treatment Completion Among Unhoused Young Adults: A Predictive Modeling Approach

Abstract

Background: Unhoused young adults aged 18–24 years are at increased risk of substance misuse relative to the general population and experience unique barriers to engaging in treatment. This study evaluates predictors of treatment completion for unhoused young adults receiving substance use treatment. Methods: Predictive models were generated on data from the 2020 Treatment Episode Data Set-Discharges. The sample included treatment discharges involving unhoused adults aged 18–24 years (N = 12,273). Model performance was assessed by inspecting several evaluative metrics. Results: Overall, each model performed relatively well (AUC: 0.7234–0.7753). Classification models trained on balanced data predicted a higher proportion of treatment completers. Models trained on balanced data also achieved higher balanced accuracy and F1 scores relative to models trained on imbalanced data. Conclusions: Findings reveal multiple features important in the accurate classification of treatment completion, which may be useful for developing individualized interventions to support clients’ engagement in treatment services.