Wearable Sensors for Exergaming Physical Exercise Monitoring via Dynamic Features

Exergaming has emerged as an innovative method to encourage physical activity through interactive video games, addressing issues such as obesity and sedentary lifestyles prevalent among various age groups. Physical inactivity and sedentary lifestyles are major health challenges for today’s youth, contributing to conditions such as obesity. In this paper we proposed an exergaming system that manipulate wearable sensors and advanced deep learning models to encourage physical engagement in a virtual reality environment. Our system enables users to play first-person physical games, promoting fitness through real-time gesture recognition. The proposed framework begins with data preprocessing, where input data is segmented using Hamming windows for consistent analysis. Next, key features are extracted using Power Spectral Density, skewness, kurtosis, and an Artificial Neural Network. Feature optimization is further enhanced through the Grey Wolf Optimization algorithm, ensuring the selection of relevant features for classification. We validated our approach on benchmark dataset ERICA, that representing a range of physical activities. Empirical findings show that our system conclude recognition accuracy of 89% for the ERICA.