This research explores the condition of substance use disorder (SUD) and patterns of substance consumption by applying three distinct unsupervised machine learning approaches aligned with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). The investigation relied on the 2019 National Survey on Drug Use and Health (NSDUH) dataset, drawn from a random sample of individuals who completed professional clinical evaluations. However, their official diagnoses remained undisclosed to the researchers. Through pattern recognition and relational insights from the models, the method supports accurate SUD classification. A licensed mental health expert thoroughly reviewed and confirmed the study’s outcomes. Specifically, the algorithms used were k-means clustering, hierarchical clustering, and density-based spatial clustering of applications with noise (DBSCAN), each applied to dedicated datasets covering alcohol, marijuana, and cocaine. Cluster evaluation incorporated silhouette scores and 95% confidence intervals (CI). Findings from this work hold potential to enhance and support conventional psychiatric assessments.
