A Precision BMI Monitoring System with Real-Time Alerts using SHAP and XGBoost based Interpretability

The aim is to develop an Enhanced Sensitivity BMI Calculator that uses machine learning, specifically XGBoost, for accurate BMI predictions and SHAP for model transparency. The system provides real-time BMI analysis and personalized health notifications to improve individual health management. Group 1 estimates body fat percentage using Random Forest Regressor, while Group 2 develops a BMI prediction model with XGBoost, using SHAP for feature selection and interpretability, and delivers real-time personalized health feedback through real-time notifications. The model achieved high accuracy with MAE = 1.22, RMSE = 1.56, and R² = 0.92, indicating strong performance in predicting BMI. SHAP analysis identified weight, age, and physical activity as the most influential factors in predictions. User feedback on the mobile app was highly positive, with ratings of 4.7/5 for accuracy and 4.8/5 for overall experience. In this work, it combines high-accuracy BMI predictions using XGBoost with transparent insights through SHAP (SHapley Additive exPlanations) analysis, offering real-time, personalized health notifications.