ARTIFICIAL INTELLIGENCE-DRIVEN EARLY PREDICTION OF CHILDHOOD OBESITY RISK USING ELECTRONIC HEALTH RECORDS AND LIFESTYLE ANALYTICS

Abstract

Childhood obesity is a growing public health issue, and its onset is complex and attributable to clinical, demographic, early life, behavioural, and lifestyle factors. Identifying those children who are at higher risk early on is key to providing early prevention interventions and minimizing the lifelong health impacts that accompany obesity. Traditional risk assessment techniques might not be effective in capturing the intricacies of interactions between several risk factors. The study aims to develop an artificial intelligence-based framework to predict the risk of childhood obesity at an early stage, combining Electronic Health Records (EHRs) with lifestyle analytics. The suggested plan will take into account regular clinical and demographic data, growth and BMI factors, health outcomes in early childhood, and lifestyle factors (physical activity level, sitting time, sleep, and nutritional intake). Data preprocessing, feature engineering, and machine learning methods are applied to build and evaluate different predictive models such as Logistic Regression, Random Forest, Support Vector Machine, XGBoost, and Neural Networks. It is suggested to use accuracy, precision, and recall along with F1-score, ROC-AUC, and calibration measures in assessing the performance of the models. Furthermore, AI methods can be used to explain the importance of features that contribute most to the prediction of obesity risk at the individual and population levels, for example via SHAP-based analysis. Overall, the proposed framework will serve as an integrated tool that will facilitate earlier identification of children at risk as well as interpretable evidence for targeted preventive interventions given the integration of healthcare and lifestyle information. The study underscores the merits of predictive analytics powered by AI in bolstering data to combat childhood obesity and individualized risk assessment.