Iot-Enabled Wearable Device For Monitoring Adolescent Physical Activity and Cardiometabolic Health

Abstract

The increasing prevalence of physical inactivity and sedentary behavior among adolescents has created a need for continuous and objective approaches for monitoring physical activity and associated cardiometabolic health indicators. This study developed an IoT-enabled wearable monitoring framework integrating multimodal movement, physiological, behavioral, and anthropometric information for automated physical-activity recognition and cardiometabolic-risk screening. The proposed wearable system incorporated a three-axis accelerometer, gyroscope, optical heart-rate sensor, oxygen-saturation sensor, and skin-temperature sensor. Sensor measurements were synchronized, preprocessed, segmented, and transformed into relevant movement and physiological features. Physical activities including sitting, standing, walking, running, and stair climbing were analyzed using a dataset comprising 893 samples with balanced representation across activity categories. Random Forest and Support Vector Machine models were employed for activity classification and cardiometabolic-risk screening. The results demonstrated clear increases in movement intensity and physiological response with increasing activity level. Accelerometer magnitude increased from 0.50 during sitting to 2.25 during running, while average heart rate increased from 85.2 BPM to 154.8 BPM. The seven-day monitoring analysis recorded an average sedentary duration of 422.4 minutes and total active duration of 228.6 minutes per day. Random Forest achieved superior activity-classification performance, with 98.3% accuracy, 98.4% precision, 98.3% recall, and a 98.3% F1-score. For cardiometabolic-risk screening, Random Forest achieved 96.8% accuracy, 96.2% sensitivity, 97.1% specificity, and an AUC of 0.982. The proposed framework demonstrated the potential of integrating wearable sensing, IoT connectivity, and machine learning for continuous adolescent activity monitoring and data-driven cardiometabolic-risk screening, providing a foundation for personalized health monitoring and preventive intervention.