Choice of optimization models for predicting spinal forces in a three-dimensional analysis of heavy work

Biomechanical models for predicting spinal compression forces are often used to assess industrial jobs with respect to the potential for low-back injury. However, there are a variety of optimization-based model formulations available for estimating internal forces from joint reaction moments in three-dimensional analyses. The question addressed in this study was, how much does the choice of model formulation affect spinal compression force estimates when analysing industrial tasks? Forty-one work postures in an aluminium reduction facility were selected for analysis. The postures were entered from videotape, and a static three-dimensional biomechanical model was used to compute the L2/L3 moments about the sagittal, frontal, and transverse planes. The shear and compression forces acting on the spine were computed from the L2/L3 moments using four optimization model formulations (minimize the sum of squared muscle stresses, sum of cubed muscle stresses, spinal compression, and a combination of maximum muscle stress and spinal compression). The anterior-posterior and compression forces predicted by the four models differed at the P < 0 . 001 level of significance. The largest difference between model predictions for a single task was 3625 N.