ai
Longitudinal and Graph-Augmented Prediction of Adolescent Substance Use Onset in the ABCD Study
- Source
- arXiv — Computers and Society
- Published
- Last verified
- 19 Aug 2026
- Confidence
- High
- Evidence
- Original document retained
- Reading time
- 1 min
- Country
- International
- Relevant to
- Research & Evidence, Technology & Data, Risk & Compliance
Executive summary
What happened, and why should leadership care?
Research conducted using data from the Adolescent Brain Cognitive Development (ABCD) Study investigated various predictive models for identifying the onset of adolescent substance use, including alcohol sipping, alcohol use, and marijuana use. The study compared cross-sectional, longitudinal, and graph-based approaches, employing tree-based models, recurrent neural networks, and Temporal Graph Convolutional Networks (T-GCNs). A key finding indicates that models incorporating longitudinal data consistently outperform those based solely on baseline characteristics.
Why this matters
Why is this strategically important?
Early identification of adolescent substance use risk is a critical public health and societal challenge. This research highlights the enhanced predictive accuracy achieved by utilizing longitudinal data and advanced modeling techniques, offering improved capabilities for proactive intervention and prevention strategies. Understanding the effectiveness of different data types and models can inform resource allocation and the development of targeted programs.
Key insights
What should be noted from the evidence?
- The study analyzed data from approximately 11,860 participants in the Adolescent Brain Cognitive Development (ABCD) Study.
- It aimed to predict the onset of alcohol sipping, alcohol use, marijuana use, and combined alcohol/marijuana use in adolescents.
- Three primary approaches were compared: cross-sectional, longitudinal, and graph-based modeling.
- Modeling techniques included tree-based models (e.g., XGBoost), recurrent neural networks, and Temporal Graph Convolutional Networks (T-GCNs).
- T-GCNs were constructed using family, school, and feature-similarity graphs to incorporate relational context.
Evidence and confidence
How far can this assessment be trusted?
High confidence. Named institution, original document retained and analysis corroborated.
Analysis is prepared editorially by Aziz Shuaib Ausi. The original publication remains the authoritative record, and executive judgement remains entirely human.
Source
Where does this originate?
Reported by arXiv — Computers and Society · International. This briefing summarises the publication for executive use; the document itself is not reproduced here.
Read the original publication