From Detection to Behavioral Change: A Scoping Review of Natural Language Processing in Substance Use Research
Year: 2026
Type: preprint
Study design: Systematic Review
Abstract: Background: Natural language processing (NLP) has emerged as a promising approach for extractingclinically relevant information from unstructured text that would otherwise be difficult to capturethrough manual review alone. Although NLP in medical settings began in the 1960s, applications haveexpanded rapidly in the 2000s, with broader use of digital narrative data such as electronic healthrecords (EHRs), clinical notes, psychotherapy transcripts, and other language-based data sources. Insubstance use research, however, the literature on NLP applications remains fragmented across datasources, methods, and clinical domains, making it difficult to identify areas of convergence andimportant gaps. Accordingly, this scoping review examined NLP applications in substance use research,focusing on clinical screening and phenotyping, risk prediction, treatment outcomes, and psychotherapyprocess.Method: A systematic search of nine clinical and technical databases identified studies publishedbetween 2016 and 2026 to examine NLP applications related to substance use behaviors, treatments,and clinically meaningful outcomes. Following screening and eligibility review, 28 studies were included.Extracted data included substance use focus, setting, data source, NLP methods, outcome domain, andkey findings.Results: Most studies (78.6%) utilized EHR clinical notes from U.S. healthcare systems, focusing primarilyon opioid outcomes. NLP methods consistently improved substance use detection compared tostructured diagnostic coding, contributing to nuanced models of overdose risk, relapse, treatmentattrition, and opioid prescribing. Recent work focused on detection and prediction to improvemultidimensional characterization and computational phenotyping of substance use populations. Acrosspsychotherapy studies, semantic, syntactic, contextual, and affective language features capturedclinically meaningful behavior processes and, in recent work, predicted subsequent alcohol use3outcomes. However, external validation performance often declined across datasets. Few studiesexamined non-opioid substances, spoken language, or community samples.Discussion: NLP methods demonstrate potential for improving substance use detection,characterization, and modeling of behaviors across clinical and behavioral domains. Current worksuggests a shift from simple classification towards richer representations of behavioral and psychosocialprocesses with notable limitations in generalizability, data diversity, and external validation remainingchallenging for substance use research.
Authors: Erin Prince, Sheridan Fine, Francesca Filbey
Language: English
Topic: Mental Health via Writing
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7214793921