Modeling the Relationship Between Electromyographic Activity and Grip Force Using AI: A Sports Biomechanics Approach

This study aimed to develop a predictive model of hand grip strength based on electromyographic (EMG) signals and to examine the variance between predicted and actual grip force values. The research involved a sample of 12 advanced-level handball players with verified medical histories. Grip strength was assessed using a customized device capable of capturing force output in Newtons at 0.1-second intervals, synchronized with EMG data recorded via the Noraxon myoMOTION system (400 Hz, 8 channels). Key EMG metrics analyzed included peak amplitude, root mean square (RMS), and time to peak activation, among others. Participants were tested at three intensity levels (50%, 75%, and 100%), each sustained for 3 seconds. Data analysis and variable selection were conducted using IBM statistical tools and an Artificial Neural Network (ANN) model specifically designed to predict grip strength based on EMG features. Findings indicated that, although the predicted grip strength values closely mirrored the actual readings, the minor differences observed were not statistically significant.