Sensitivity of kinematics-based model predictions to optimization criteria in static lifting tasks

The effect of eight different cost functions on trunk muscle forces, spinal loads and stability was investigated. Kinematics-based approach combined with nonlinear finite element modeling and optimization were used to model in vivo measurements on isometric forward flexions at ∼40° and ∼65° in sagittal plane with or without a load of 180 N in hands. Four nonlinear (∑stress3, ∑stress2, ∑force2 and muscle fatigue) and four linear (∑stress, ∑force, axial compression and double-linear) criteria were considered. Predicted muscle activities were compared with measured EMG data. All predictions, irrespective of the cost function used, satisfied required kinetic, kinematics and stability conditions all along the spine. Four criteria (∑stress3, ∑stress2, fatigue and double-linear) predicted muscle activities that qualitatively matched measured EMG data. The fatigue and double-linear criteria were inadequate in predicting greater forces in larger muscles with no consideration for their moment arms. Nearly the same stability margin was computed under these four cost functions. At the lower lumbar levels, the compression forces differed by <20% and the shear forces by <14% as various cost functions were considered. Smaller axial compression and anterior shear forces (by ⪝6%) were computed when only the active components rather than the total muscle forces were taken as unknown in the ∑stress3 cost function. Overall, one single cost function of ∑stress2 or ∑stress3 rather than a multi-criteria one was found sufficient and adequate in yielding plausible results comparable with measured EMG activities and disc pressure.