Estimation of Electrically-Evoked Knee Torque from Mechanomyography Using Support Vector Regression

Abstract

The difficulty of real-time muscle force or joint torque estimation during neuromuscular electrical stimulation (NMES) in physical therapy and exercise science has motivated recent research interest in torque estimation from other muscle characteristics. This study investigated the accuracy of a computational intelligence technique for estimating NMES-evoked knee extension torque based on the Mechanomyographic signals (MMG) of contracting muscles that were recorded from eight healthy males. Simulation of the knee torque was modelled via Support Vector Regression (SVR) due to its good generalization ability in related fields. Inputs to the proposed model were MMG amplitude characteristics, the level of electrical stimulation or contraction intensity, and knee angle. Gaussian kernel function, as well as its optimal parameters were identified with the best performance measure and were applied as the SVR kernel function to build an effective knee torque estimation model. To train and test the model, the data were partitioned into training (70%) and testing (30%) subsets, respectively. The SVR estimation accuracy, based on the coefficient of determination (R2) between the actual and the estimated torque values was up to 94% and 89% during the training and testing cases, with root mean square errors (RMSE) of 9.48 and 12.95, respectively. The knee torque estimations obtained using SVR modelling agreed well with the experimental data from an isokinetic dynamometer. These findings support the realization of a closed-loop NMES system for functional tasks using MMG as the feedback signal source and an SVR algorithm for joint torque estimation.

Electrically evoked wrist extensor muscle fatigue throughout repetitive motion as measured by mechanomyography and near-infrared spectroscopy

“…Few studies had verified that MMG is a reliable tool to measure the development of muscle fatigue when torque output is not, or hardly, detectable such as during FES-evoked rehabilitation for individuals with paralysis or paresis [7,30]. It has been well documented that the amplitude and frequency of MMG signal are both related to muscle force [30,31] and torque in the SCI population population during electrical stimulation [32,33]. Muscle fatigue from lower back pain using EMG, MMG and NIRS has also been investigated, and the authors concluded that a simultaneous recording system of multiple signals might be a more promising approach with regards to quantifying muscle fatigue [34] and such system that could predict and detect onset and rate of fatigue might be clinically efficacious for rehabilitation purposes.…”

Assessment of muscle activity using electrical stimulation and mechanomyography: a systematic review

“…The findings of these studies have supported the use of muscle mechanics, anthropometry, and age difference in terms of characterizing the muscle torque. Machine learning [ 83 , 86 ] was also determined to provide greater results which support the hypothesis that sophisticated signal processing may provide promising future improvements in terms of MMG-based muscles studies.…”

“…A study conducted a year later [ 83 ] used computational intelligence to solve the complex mathematical computation derived in the above-mentioned previous studies [ 82 ]. Using , the knee angle, and the levels of NMES parameters, the SVR was trained to predict torque; then, the knee torque was in line with the dynamometer readings.…”

“…The high correlation and reduced root mean square error only confirmed the efficiency of the developed technique in terms of estimating the knee torque and the accuracy of the laboratory results. Similarly, the variability in the neurological function of SCI patients determined in this study [ 83 ] was addressed using artificial intelligence, which was performed in a subsequent study [ 86 ]. After delivering electrical impulses to the quadriceps muscle groups of three SCI subjects, the knee torque estimated using a dynamometer was significantly correlated with the ANN-predicted torque.…”

Mechanomyography and Torque during FES-Evoked Muscle Contractions to Fatigue in Individuals with Spinal Cord Injury

“…The condition of rapid-onset muscle fatigue may even lead to muscle damage [ 22 ]. Even though there are other muscle fatigue monitoring techniques including dynamometer, EMG, NIRS and other force sensors currently available, the MC sensor may be a practical choice for clinicians due to its reliability and ease of use compared to other techniques [ 23 , 24 ]. Such a use of an MC sensor has been previously proposed by Dordevic and colleagues [ 16 ], who demonstrated a strong correlation between MC voltage and force in 21 able-bodied subjects (R 2 = 85%).…”