Evaluating Different Optimization Criteria for Estimating Spine Loads and Muscle Activity During Manual Lifting With...

We evaluated how back muscle activity, estimated using different optimization criteria in a commercial musculoskeletal modeling system, corresponds to measured activity. We also examined how the choice of optimization criterion influences the effect of back-support exoskeleton (BSE) on reducing spine loads during symmetric and asymmetric dynamic lowering/lifting tasks.

We simulated dynamic lowering/lifting tasks in the AnyBody Modeling System™ using three optimization criteria: quadratic, cubic, and min/max, across three intervention conditions: two BSEs and one control (no BSE). Estimated muscle activity was compared with normalized electromyography (nEMG) using quantitative and qualitative assessments. For each simulation, we estimated peak axial compression and anteroposterior shear spine loads at the L4/L5 level.

Quadratic and cubic criteria showed stronger associations with nEMG and smaller errors. Errors were larger during lifting than lowering and greater in asymmetric than symmetric tasks. Qualitatively, the quadratic and cubic estimated higher ipsilateral than contralateral longissimus erector spinae activity in asymmetric tasks, contrary to measured patterns. EMG showed lower activity during lowering than lifting, whereas all criteria produced nearly symmetrical phase profiles. Estimated spine loads varied by criterion; however, all indicated comparable relative reductions with BSE use.

Optimization criteria influenced the magnitude of estimated spinal loads but did not affect conclusions regarding BSE effectiveness. Although muscle activity estimates showed generally strong agreement with measured data, consistent discrepancies in asymmetric tasks indicate limitations of current static optimization approaches in capturing true muscle recruitment patterns. Criterion performance was not substantially altered by BSE use, suggesting task dependency rather than device dependency.