Comparison of trunk muscle forces, spinal loads and stability estimated by one stability- and three EMG-assisted optimization approaches

• Predicted muscle/spinal forces by three EMG and an optimization models were compared.
• Substantial differences in predicted muscle forces, but not spinal loads, were found.
• Large range of gain values was estimated in each EMG model during lifting tasks.

Various hybrid EMG-assisted optimization (EMGAO) approaches are commonly used to estimate muscle forces and joint loads of human musculoskeletal systems. Use of EMG data and optimization enables the EMGAO models to account for inter- and intra-individual variations in muscle recruitments while satisfying equilibrium requirements. Due to implications in ergonomics/prevention and rehabilitation/treatment managements of low-back disorders, there is a need to evaluate existing approaches. The present study aimed to compare predictions of three different EMGAO and one stability-based optimization (OPT) approaches for trunk muscle forces, spinal loads, and stability. Identical measured kinematics/EMG data and anatomical model were used in all approaches when simulating several sagittally symmetric static activities. Results indicated substantial inter-model differences in predicted muscle forces (up to 123% and 90% for total muscle forces in tasks with upright and flexed postures, respectively) and spinal loads (up to 74% and 78% for compression loads in upright and flexed postures, respectively). Results of EMGAO models markedly varied depending on the manner in which correction (gain) factors were introduced. Large range of gain values (from ∼0.47 to 41) was estimated in each model. While EMGAO methods predicted an unstable spine for some tasks, OPT predicted, as intended, either a meta-stable or stable states in all simulated tasks. An unrealistic unstable state of the spine predicted by EMGAO methods for some of the simulated tasks (which are in reality stable) could be an indication of the shortcoming of these models in proper prediction of muscle forces.