A Muscle/Ligament Model to Predict Loads on L5/S1 during Three Dimensional Lifting

Abstract

A muscle / ligament model was constructed to estimate the spinal load that resulted from both muscle and ligament sources during dynamic, three dimensional lifting tasks. The model was tested using data from ten subjects performing lifts over a range of realistic industrial conditions (velocity: 10, 20, or 30 degrees / second; asymmetry: 0, 15, or 30 degrees; and weight lifted: 13.6 or 22.7 Kg). During the task, three dimensional trunk position, trunk velocity, the reaction forces, the reaction moments, and the electromyography of the major trunk musculature were collected. Ligaments were represented in the model as vectors spanning the lumbar spine, with their stress—strain properties taken from the literature. The muscle components were modeled based on the OSU Biodynamic EMG assisted model but excluded any effect not resulting solely from active force generation. Thus, the trunk muscles were also represented by vectors spanning the lumbar spine. For each subject, the model was calibrated for both muscle and ligament moment generation by comparing the predicted moment to the measured applied moment in regions where the appropriate moment component has been shown to dominate.

The muscle / ligament model was found to predict the moment at L5/S1 at least as accurately as the muscle—only model (the previously reported OSU Biodynamic EMG assisted model which indirectly combines ligament effects into the muscle effects). Both models predicted moment with an average R square value of 0.8, and average error of 23 N m (p > 0.05). For symmetric upright postures, there was no influence of ligaments so both models predicted similar compression of about 1675 N m. As asymmetry or flexion angle increased, the muscle / ligament model predicted higher compression as a result of the smaller moment arms of the ligaments. In the most extreme posture (40 degrees flexion with 30 degrees asymmetry), the predicted compression from the muscle / ligament model (4250 N) was significantly larger than the muscle—only model (3680N). Finally, all asymmetric conditions resulted in predictions from the muscle / ligament model that exceeded the NIOSH 3400 N tolerance, but only the most asymmetric condition resulted in predictions from the muscle—only model that exceeded the NIOSH limit. Thus, muscle models that do not account for ligament effects may be ineffective in accurately evaluating compression during certain job tasks.