Estimation of hip abduction moment based on body fixed sensors

Abstract

Background. During standing and walking, abnormal trunk posture or movements may result from insufficient hip abductor force. Quantifying abduction moments requires inverse dynamics based on an appropriate model and measured data. Contrary to using a force plate and motion analysis system, body fixed sensors permit an analysis over subsequent strides and in a natural environment. Hence, the present paper analyses the feasibility of body fixed sensor-based estimations of the hip abduction moment. Methods. Two models were defined to estimate the hip abduction moment based on data measured by accelerometers and gyroscopes; one model assumes a rigid trunk, the other assumes an upper and lower trunk segment. Data of five healthy subjects in different loading conditions while standing or walking were used to compare sensor-based estimations with estimations based on measured position and force data. Findings. In different loading conditions, mean abduction moments calculated from position and force data varied between 0.27 and 0.62 N m/kg. Generally, both sensor-based methods overestimated abduction moments. Results of a segmented trunk model showed best correspondence to the position and force-based method. Individual correlation coefficients between mean abduction moments in different loading conditions ranged from 0.72 to 0.98 (rigid trunk model) and 0.89 to 0.95 (segmented trunk model). Interpretation. Hip abduction moments can be estimated based on the two models that use sensor data. Both methods are sensitive to different loading conditions. However, estimations are only valid when the trunk is supported by one leg alone, and particularly when the rigid trunk assumption is violated, the segmented trunk model yields better results.