AUTHOR=Helforoush Zarindokht , Sayyad Hossein 

TITLE=Prediction and classification of obesity risk based on a hybrid metaheuristic machine learning approach

JOURNAL=Frontiers in Big Data

VOLUME=Volume 7 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/big-data/articles/10.3389/fdata.2024.1469981

DOI=10.3389/fdata.2024.1469981

ISSN=2624-909X

ABSTRACT=As the number of obese people grows, it becomes a major global health problem that needs new ways to accurately predict and tackle. Even though traditional regression models are useful, they often fail to show how genetic, environmental, and behavioral factors interact to cause obesity. In response, this study investigates the potential of machine-learning methodologies to enhance obesity risk prediction. Through meticulous data preprocessing and rigorous evaluation of various supervised learning algorithms, including the pioneering ANN-PSO hybrid model, remarkable progress is achieved in forecasting obesity risk levels. With its noteworthy accuracy rate of 92%, the proposed model outperforms traditional regression approaches, underscoring the transformative impact of advanced machine-learning techniques in public health research and practice. Additionally, SHAP was employed as a method for feature importance analysis to evaluate the effect of different features on obesity classes, providing deeper insights into the factors contributing to obesity risk. By providing nuanced insights into obesity risk profiles, our model offers a promising avenue for personalized healthcare interventions, empowering healthcare providers to tailor preventive measures and treatment strategies to individual needs. These findings underscore the imperative of integrating innovative machine-learning methodologies into public health initiatives to address the escalating obesity epidemic on a global scale effectively.