Intelligent integrated wearable sensing mechanism for vertical jump height prediction in female netball players

Abstract

Vertical jump (VJ) height is a fundamental performance analysis parameter in several sports involving frequent jump-landing maneuvers such as netball. Recent studies have largely examined performance parameters associated with vertical jump height (VJH) in isolation from each other. This study presents an investigation into the relationship between integrated performance parameters (IPP) and VJH during single-leg (VJSL) and double-leg (VJDL) vertical jump tests. IPP considered include; electromyography (EMG) activity of eight lower extremity (LE) muscles, 3D Kinematics of the knee and ankle joints, body height (BH), reach height (RH), and Jump duration (JT). Thirteen healthy national female netball players participated in this study. Each subject performed VJSL and VJDL in three trials while simultaneously and synchronously recording their LE-EMG activity, 3D kinematics, and VJH in each jump trial. LE-EMG activity acquisition was through wirelessly transmitting BioRadio units (CleveMed Inc. USA), while 3D kinematics were obtained through a six-3D marker-based motion capture camera system (Qualisys Inc. Sweden). VJH reading was obtained from a vertec device (Power Systems Inc. USA). A total of 22 IPP were extracted from raw data of both VJSL tests (i.e VJSL Right-Leg (VJSLR), and VJSL Left-Leg (VJSLL)), while 44 IPP were extracted from raw data of VJDL. The relationship between the reduced datasets' parameters and response variable (VJH) was then modeled using Multilayer Perceptron Feed Forward Neural Networks (FFNNs). Significant features were further selected through stepwise regression analysis. Results showed that FFNNs trained with Scaled conjugate gradient back-propagation (SCG) algorithm achieved the best VJH prediction with accuracy of 97.39% for VJSLL, 94.52% for VJSLR, and of 96.74% for VJDL. These results demonstrate that the integration of 3D Kinematics and EMG using wearable sensors interfaced with motion capture system for IPP, has led to more accurate prediction of VJH. Thus, this serves as quantifiable feedback to coaches and players for performance enhancement as well as injury prevention in jump landing tasks investigated.