Influence of input data errors on the inverse dynamics analysis of human locomotion

Abstract

Inverse dynamic techniques are used in gait analysis to calculate the net joint moments that the musculoskeletal system produces during human locomotion. Errors present in the input data may affect significantly the results of the inverse dynamics problem. In this work, the influence of inaccuracies in body segment parameters (BSP) and ground reaction force measurements on the obtained results is analyzed. The uncertainties in these data are computer-generated as perturbations added to their actual values. On the one hand, the uncertainties in BSP are modelled as statistical errors using zero-mean Gaussian distributions. On the other hand, inaccurate ground reaction forces are represented adding a percentage of error to the theoretical force. Errors present in the ground reaction forces allow to estimate the uncertainty introduced to the analysis when ambiguous force plate measurements are used as input parameters. The paper presents detailed results and discussions to quantify the effects of the above errors on the inverse dynamics analysis. The several simulations carried out show that the results are more sensitive to errors in the ground contact forces –when these are used– than in the body inertial parameters.