Influence of Joint Angle on EMG-Torque Model During Constant-Posture, Torque-Varying Contractions

Relating the electromyogram (EMG) to joint torque is useful in various application areas, including prosthesis control, ergonomics and clinical biomechanics. Limited study has related EMG to torque across varied joint angles, particularly when subjects performed force-varying contractions or when optimized modeling methods were utilized. We related the biceps-triceps surface EMG of 22 subjects to elbow torque at six joint angles (spanning 60 ° to 135°) during constant-posture, torque-varying contractions. Three nonlinear EMG σ-torque models, advanced EMG amplitude (EMG σ) estimation processors (i.e., whitened, multiple-channel) and the duration of data used to train models were investigated. When EMG-torque models were formed separately for each of the six distinct joint angles, a minimum “gold standard” error of 4.01±1.2% MVC F90 resulted (i.e., error relative to maximum voluntary contraction at 90 ° flexion). This model structure, however, did not directly facilitate interpolation across angles. The best model which did so achieved a statistically equivalent error of 4.06±1.2% MVC F90. Results demonstrated that advanced EMG σ processors lead to improved joint torque estimation as do longer model training durations.