ENHANCING TRUST IN OBESITY PREDICTION USING EXPLAINABLE STACKING MODELS

Abstract

The accuracy and reliability of obesity prediction are enhanced by the explainable stacking-based ensemble learning approach employed in this study, which integrates machine learning models. For the prevention of obesity, a global health concern, it is necessary to conduct an early and precise risk assessment. Decision trees, SVMs, and logistic regression comprise fundamental learners. Meta-learners enhance their performance by integrating predictions. Ensemble models are elucidated by Explainable AI (XAI). The decision patterns, feature contributions, and model outputs are all described by these methods. Interpretability and predictive power are enhanced by specifying BMI, nutrition, exercise, and lifestyle. The explainable layering model outperforms single classifiers and enhances the confidence and utility of clinical and public health. Experimental results support this assertion.