Evaluating dynamic similarity of fixed, self-selected and anatomically scaled speeds in non-linear analysis of gait during treadmill running

The aim of this study is to evaluate how speed affects non-linear measures of variability. Fixed and self-selected speeds were compared to an anatomically scaled speed calculated based on leg length to evaluate which provided a more reproducible result between subjects. Sixteen subjects ran on a treadmill at a fixed, scaled and self-selected speed and at ±10% in each case. Kinematic data were collected for two minutes at 250 Hz for each trial. Sample entropy (SaEn) and maximum Lyapunov exponents (LyE) were calculated from the sagittal knee and hip joint angles to evaluate regularity of gait and local stability. These nonlinear measures were compared to evaluate the dynamic similarity of the movement in each case, and to evaluate speed as a confounding variable in non-linear analysis. An anatomically scaled speed shows more dynamic similarity than a fixed or self-selected speed with the lowest observed coefficient of variation for each measure. This was found to be statistically significant for both nonlinear measures of the hip (SaEn p = 0.038; LyE p = 0.040). Speed was not found to be a confounding variable in non-linear analysis of running gait of a healthy population (η2