HYBRID TREE-ENSEMBLE MODELS INTEGRATING EXTREME VALUE THEORY FOR OBESITY ANALYSIS IN SAUDI ARABIA
JP Journal of Biostatistics · 2025
DOI: 10.17654/0973514325026 ↗Kurz gefasst
Fachartikel in der Zeitschrift JP Journal of Biostatistics, 2025. Verfasst von Wadha Mohammed T. Alanazi und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Obesity represents a significant public health crisis in Saudi Arabia, exacerbated by rapid urbanization, dietary shifts, and sedentary lifestyles. Traditional statistical methods often inadequately address the complex, non-linear relationships among obesity determinants. This study proposes a novel machine learning framework integrating hybrid tree-ensemble models - Random Forest (RF) and Gradient Boosting Machine (GBM) - with SHapley Additive exPlanations (SHAP) for interpretability and Extreme Value Theory (EVT) for outlier analysis. Using a nationally representative dataset, we trained and validated models to identify key predictors of obesity (BMI 30) and assess extreme-risk cases. The EVT-augmented hybrid model achieved superior performance (accuracy: 86.2%, MSE: 0.20) compared to baseline RF (82.5%) and GBM (84.3%) models. SHAP analysis revealed BMI age, and physical inactivity as dominant predictors, while EVT quantified tail risks (shape parameter in severe obesity. Our approach demonstrates that machine learning, combined with interpretability techniques, can effectively disentangle multifactorial obesity drivers and support targeted interventions. These findings provide a methodological advancement in obesity analytics and offer evidence-based insights for public health policy in Saudi Arabia.
Zitieren
APA
Alanazi, W. M. T., & Bakar, M. A. A. (2025). HYBRID TREE-ENSEMBLE MODELS INTEGRATING EXTREME VALUE THEORY FOR OBESITY ANALYSIS IN SAUDI ARABIA. JP Journal of Biostatistics. https://doi.org/10.17654/0973514325026
Vancouver
Alanazi WMT, Bakar MAA. HYBRID TREE-ENSEMBLE MODELS INTEGRATING EXTREME VALUE THEORY FOR OBESITY ANALYSIS IN SAUDI ARABIA. JP Journal of Biostatistics. 2025. doi:10.17654/0973514325026
BibTeX
@article{Alanazi2025,
author = {Alanazi, Wadha Mohammed T. and Bakar, Mohd Aftar Abu},
title = {HYBRID TREE-ENSEMBLE MODELS INTEGRATING EXTREME VALUE THEORY FOR OBESITY ANALYSIS IN SAUDI ARABIA},
journal = {JP Journal of Biostatistics},
year = {2025},
doi = {10.17654/0973514325026}
}
EUVIMED