Smart Fitness Insights: Predicting Exercise Calories with Explainable AI

Abstract: Accurate estimation of energy expenditure and calories burned during exercise is essential for fitness tracking and health monitoring. Reliable calorie estimation enables professionals to design personalized fitness plans and helps individuals optimize their workouts. This study proposes a machine learning approach to predict calories burned based on physiological and exercise-related features such as gender, age, height, weight, exercise duration, heart rate, and body temperature. Several ensemble regression models are employed, including Gradient Boosting Decision Trees Regression (GBDTR), Extreme Gradient Boosting Regression (XGBOOSTR), Stacking Regression (STACKINGR), Random Forest Regression (RFR), Bagging Regression (BAGGINGR), and Voting Regression (VOTINGR). Among these models, XGBOOSTR demonstrates the highest performance with a Mean Squared Error (MSE) of 14.224, Mean Absolute Error (MAE) of 2.022, R-squared (R²) value of 0.9964, Peak Signal to Noise Ratio (PSNR) of 37.41, and Signal to Noise Ratio (SNR) of 29.29. Explainable Artificial Intelligence (XAI) techniques, including Local Interpretable Model Agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP), are applied to interpret model predictions and identify the most influential features, such as exercise duration, heart rate, and body temperature. The findings of this research provide valuable insights for developing wearable health applications, enhancing personalized fitness tracking, and assisting medical professionals in promoting healthier lifestyles.

Keywords: Machine Learning, XAI, Regression, Prediction.