Submaximal electromyography-driven musculoskeletal modeling of the human trunk during static tasks: Equilibrium and...

Conventional electromyography-driven (EMG) musculoskeletal models are calibrated during maximum voluntary contraction (MVC) tasks, but individuals with low back pain cannot perform unbiased MVCs. To address this issue, EMG-driven models can be calibrated in submaximal tasks. However, the effects of maximal (when data points include the maximum contraction) and submaximal calibration techniques on model outputs (e.g., muscle forces, spinal loads) remain yet unknown. We calibrated a subject-specific EMG-driven model, using maximal/submaximal isometric contractions, and simulated different independent tasks. Both approaches satisfactorily predicted external moments (Pearson's correlation ∼ 0.75; relative error = 44%), and removing calibration tasks under axial torques markedly improved the model performance (Pearson's correlation ∼ 0.92; relative error ∼ 28%). Unlike individual muscle forces, gross (aggregate) model outputs (i.e., spinal loads, stability index, and sum of abdominal/back muscle forces) estimated from maximal and submaximal calibration techniques were highly correlated (r > 0.78). Submaximal calibration method overestimated spinal loads (6% in average) and abdominal muscle forces (11% in average). Individual muscle forces estimated from maximal and submaximal approaches were substantially different; however, gross model outputs (especially internal loads and stability index) remained highly correlated with small to moderate relative differences; therefore, the submaximal calibration technique can be considered as an alternative to the conventional maximal calibration approach.