Machine Learning-Based Classification of Body Imbalance and Its Intensity Using Electromyogram and Ground Reaction Force in Immersive Environments.

Autor: Gul, Jahan Zeb, Cheema, Muhammad Omar, Mohy Ud Din, Zia, Khan, Maryam, Kim, Woo Young, Rehman, Muhammad Muqeet
Předmět:
Zdroj: Applied Sciences (2076-3417); Sep2024, Vol. 14 Issue 18, p8209, 17p
Abstrakt: Body balancing is a complex task that includes the coordination of muscles, tendons, bones, ears, eyes, and the brain. Imbalance or disequilibrium is the inability to maintain the center of gravity. Perpetuating body balance plays an important role in preventing us from falling or swaying. Biomechanical tests and video analysis can be performed to analyze body imbalance. The musculoskeletal system is one of the fundamental systems by which our balance or equilibrium is sustained and our upright posture is maintained. Electromyogram (EMG) and ground reaction force (GRF) monitoring can be utilized in cases where a rapid response to body imbalance is a necessity. Body balance also depends on visual stimuli that can be either real or virtual. Researchers have used virtual reality (VR) to predict motion sickness and analyze heart rate variability, as well as in rehabilitation. VR can also be used to induce body imbalance in a controlled way. In this research, body imbalance was induced in a controlled way by playing an Oculus game and, simultaneously, EMG and GRF were recorded. Features were extracted from the EMG and were then fed to a machine learning algorithm. Several machine learning algorithms were tested and upon 10-fold cross-validation; a minimum accuracy of 71% and maximum accuracy of 98% were achieved by Gaussian Naïve Bayes and Gradient Boosting classifiers, respectively, in the classification of imbalance and its intensities. This research can be incorporated into various rehabilitative and therapeutic systems. [ABSTRACT FROM AUTHOR]
Databáze: Complementary Index