+ HEALTH Kidney stones (KS) are a common urological condition, the aetiology of which remains incompletely understood.…± HEALTH Background Optimizing adjuvant chemotherapy (AC) for gastric cancer (GC) remains challenging due to patient h…+ HEALTH Background Hepatocellular carcinoma (HCC) is characterized by marked heterogeneity and an immunosuppressive m…+ HEALTH Objectives To identify risk and protective factors associated with lifetime marijuana use among male high sch…± EDUCATION Introduction The integration of artificial intelligence (AI) tools into radiation therapy workflows offers si…± HEALTH Background Prognosis remains heterogeneous among critically ill patients with acute kidney injury (AKI). We e…± CLIMATE Accurate estimation of aboveground biomass (AGB) is essential for sustainable forest management, carbon accou…+ HEALTH Objectives Cervical cancer is a leading female malignancy with high global morbidity/mortality, and remains h…
TruaceTracing the truth around AIMonday, August 24, 2026
Subscribe
Health·G Space·Evidence-backed gain·Published 2026-08-24

Identifying risk factors for marijuana use among male high school students using machine learning: Implications for public health

Abstract: Objectives To identify risk and protective factors associated with lifetime marijuana use among male high school students through an interpretable machine learning model, providing evidence to support early and targeted public health interventions. Study design Cross-sectional analysis of 2023 Youth Risk Behavior Surveillance System (YRBS) data for boys in grades 9-12 across the United States. Methods The final analytical sample included 8285 boys after excluding missing outcomes. Thirty-six predictors spanning…

TRV-2026-0864Peer-reviewedPermanent record — cite & verify
Identifying risk factors for marijuana use among male high school students using machine learning: Implications for public health

"Public Health Lab Ribbon Cutting Ceremony" by MDGovpics is licensed under CC BY 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/2.0/.

The quick read

A cross-sectional study of 8285 male high school students from the 2023 Youth Risk Behavior Surveillance System used 36 demographic, behavioral, and psychosocial predictors to model lifetime marijuana use, reported by 28.4% of participants. After benchmarking 17 algorithms, an optimized logistic regression model with SHAP and LIME explanations achieved AUC 0.9034 and accuracy 0.8582.

The work matters because it translates interpretable machine learning into actionable public health signals, highlighting co-occurring substance use, social media use, age, grade, bullying and sexual violence exposure, and academic achievement as key factors. As of the August 2026 publication date, results remain associational from a single US survey of boys, leaving uncertainty about causality, generalizability to other groups, and effectiveness of subsequent interventions.

Main points
  • Analysis used 2023 Youth Risk Behavior Surveillance System data for 8285 boys in grades 9-12 after excluding missing outcomes.
  • Thirty-six predictors spanning demographics, substance use, lifestyle, psychosocial stressors, mental health, and household environment were retained.
  • Seventeen algorithms were benchmarked; Logistic Regression was selected for predictive performance and interpretability, with SHAP and LIME for explanation.
  • Lifetime marijuana use was reported by 28.4% of participants; top predictors included electronic vapor product use, cigarette smoking, alcohol consumption, and frequent social media use.
Gain

An interpretable logistic regression model trained on YRBS data predicted lifetime marijuana use among male high school students with high accuracy, enabling early identification of at-risk boys for targeted school and community prevention.

The rundown

Researchers analyzed 2023 YRBS data for boys in grades 9-12, retaining 36 predictors after optimizing missing data thresholds and determining ensemble feature importance using Logistic Regression, Linear Discriminant Analysis, and Extra Trees classifiers.

The optimized logistic regression model was well calibrated with Brier score = 0.104 and expected-to-observed ratio = 1.01, and SHAP and LIME analyses confirmed robustness; high academic achievement was found to be protective while bullying, sexual violence exposure, and other substance use were predictive.

Sources

Reader signal

How should this claim be treated?

The debate