Body Fat Percentage Detection Using Machine Learning

This study presents a machine leaming-based approach for predicting body fat percentage and identifying associated health risks using dietary intake patterns and physical attributes. The primary objective of the work is to develop an accessible and non-invasive method that eliminates the need for specialized medical. equipment. The system collects anthropometric measurements such as height, weight, waist, hip, and forearm length, along with daily meal-level food entries including breakfast, lunch, and dinner items. A Linear Regression model is employed to estimate the body fat percentage based on engineered features derived from both physical and dietary data. Subsequently, an XGBoost classifier analyzes the predicted body fat percentage to detect potential diseases and risk levels related to obesity, such as diabetes and cardiovascular conditions. Experimental evaluation demonstrates that the model achieves consistent prediction accuracy and effectively classifies health risks, offering a practical decision-support tool for early health assessment. The findings Indicate that integrating real-world diet patterns with physical measurements provides meaningful insights for personalized health monitoring. Overall, the system contributes to improved awareness of body composition, highlights risk factors linked to unhealthy body fat levels, and supports preventive healthcare through data-driven analysis.