Coordination and postural control using multivariate analysis

Abstract

The aim of this paper is to run the independent component analysis for human movement studies. The subjects were 8 women, who should stand still and raise their arms, holding a 2kg mass weight with their hands. Eight muscles were monitored with electromyography (EMG), and shoulder flexion and extension were measured with a electrogoniometer, therefore, their temporal data, during anticipatory (APA) and compensatory (APC) postural adjustments, were analyzed using principal component and independent component analyses. We run multiple linear regression for 2 and 3 independent components, 2D e 3D, respectively, from EMG. For 2D e 3D, the regression residuals were higher during APC e high conditions. For APA, the residuals were higher during 3D and high condition. For APC, the residuals were higher during 2D and high condition. We observe that the same postural perturbation may lead to different control responses by the variation of principal components.