Variability in spine loading model performance q

Objective. To assess the sources of variability associated with an EMG-assisted model of spine loading. Design. In vivo measurements of trunk dynamics, lifting moments and muscle activities were used as inputs into an EMG-assisted spine loading model. Background. Several types of variability are inherent in biomechanical assessments of risk associated with trunk bending motions during lifting. Variability may occur as a function of variations in spine loading due to either subject variations in motion pro®les (kinematics) or biomechanical model performance. Methods. Twelve experienced and inexperienced materials handlers performed 10 repeated lifts where load weight, asymmetry, and velocity were varied. The experiment was replicated on a second day to assess day to day variability. Results. These model performance variables indicated that variability was mainly a function of subject characteristics and experience. Minor variations in variability were associated with the task asymmetry and weight lifted. Advanced analyses suggested that experienced workers had a greater range of back motion compared to inexperienced workers which would a€ect the length± strength component of the model calibration. Conclusions. This study indicates that for the results of an EMG-assisted model to be accurate, it is important to ensure that the model re ̄ects a realistic relationship between the trunk muscle length and the muscle force production capacity. Underestimation if this relationship can degrade model ®delity and robustness. Relevance These results imply that by properly calibrating the model it is then reasonable to assume that the vast majority of variations observed in repeated exertions of a particular trial are due to kinematic and kinetic di€erences inherent in the muscle control system and not a function of model randomness. Ó 1999 Elsevier Science Ltd. All rights reserved.