Truace
Subscribe

Broader factors associated with current adolescent overweight/obesity status in YRBSS 2023: Survey-aware analysis and calibrated AI-assisted classification

Adolescent overweight and obesity are shaped by demographic, behavioral and psychosocial factors, but it remains unclear whether broader predictor information improves cross-sectional classification of current overweight/obesity status more consistently than changing the machine-learning algorithm. This cross-sectional secondary analysis used 2023 Youth Risk Behavior Surveillance System data from 17,814 United States adolescents aged 12-18 years. Survey-aware multivariable logistic regression was used to identif…

By the Truvace desk2 min read1 sourcePeer-reviewedTRV-2026-1709· cite & verify
Broader factors associated with current adolescent overweight/obesity status in YRBSS 2023: Survey-aware analysis and calibrated AI-assisted classification
"Fulton parents confront school board about controversial survey" by KOMUnews is licensed under CC BY 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/2.0/.

In brief

Researchers analysed 2023 Youth Risk Behavior Surveillance System responses from 17,814 US adolescents aged 12-18. They compared Random Forest and XGBoost models built on base versus extended predictor sets, alongside survey-weighted logistic regression, to classify current overweight or obesity.

The results suggest that richer predictor information helped more consistently than the choice of algorithm, and that recalibration made predicted risks more trustworthy. Even so, discrimination stayed modest, and the models describe current status rather than future obesity, so any screening use remains unproven until validated in new settings.

Main points

  1. Overweight/obesity prevalence in the sample was 31.8%.
  2. Male sex and Black or African American and Hispanic/Latino race/ethnicity were linked to higher odds of overweight/obesity.
  3. Higher moderate-to-vigorous physical activity and more frequent breakfast intake were linked to lower odds.
  4. Isotonic recalibration cut expected calibration error from 0.1160 to 0.0213 while preserving discrimination.
  5. Exploratory SHAP analysis ranked sex, race/ethnicity, physical activity, sleep duration, age and breakfast frequency among the strongest contributors.

The gain

Extended predictor sets raised discrimination for both tree-based models on held-out data, and isotonic recalibration made predicted probabilities far more reliable for the selected Random Forest model.

The rundown

The study is a cross-sectional secondary analysis. Survey-aware multivariable logistic regression identified independent correlates, while Random Forest and XGBoost were each tested under base and extended predictor frameworks.

Models were assessed on a held-out test set for discrimination, threshold-dependent performance, calibration and interpretability. RF-Extended was chosen for calibration and interpretation, and SHAP was used for an exploratory look at which features drove its output.

The authors state that predictors and BMI status came from the same survey cycle, so the models classify current status and should not be read as predicting future obesity. They also mention possible use among Vietnamese adolescents but say this would need prospective and external validation.

Sources

  1. Peer-reviewedPLOS One2026-10-08

The debate