Principal motion ellipsoids: Gait variability index based on principal motion analysis

We developed a gait variability index on the basis of principal motion analysis (PMA). PMA decomposes multivariate time-series data into independent bases. The resulting variability index evaluates how individual gaits vary along these bases. Hence, it allows us to analyze how and to what extent the gait motions of humans vary, unlike previous popular indices, which aimed to quantity the level of gait variability. We computed this variability index for lower limb joints and the velocity of the center of mass during walking. These indices exhibited higher correlation coefficients with MeanSD, which is a typical gait variability index. The meanings of the independent bases was successfully interpreted, suggesting the validity and usability of the new gait variability index.