Motor-equivalent covariation stabilizes step parameters and center of mass position during treadmill walking

Abstract

We investigated motor-equivalent stabilization of task-related variables (TRV) at times of heel strike in eight healthy young men (23-30 years) who walked on a motorized treadmill at self-selected and prescribed speeds within the normal walking speed range. The TRV consisted of step parameters (step length and width) and the center of mass (CoM) position relative to the support (back and front feet). Motor-equivalent stabilization of the TRV was assessed using a decorrelation technique, comparing empirical to decorrelated (covariation-free) variability. Analysis indicated reliable covariation for all TRV. In both the fore-aft and lateral directions, stabilization by covariation was highest for CoM position relative to the front foot, indicating a prioritization of equilibrium-related variables. Correlations among TRV revealed that the relation between CoM and step parameter control differed between the fore-aft and lateral directions. While stabilization of lateral foot position appears to be due to control of CoM relative to each foot, step length showed small, but reliable, stabilization beyond CoM stabilization, which may be related to spatiotemporal regularity of the step pattern.

Analysis of gait within the uncontrolled manifold hypothesis: Stabilisation of the centre of mass during gait

“…Only two (Black et al, 2007;Verrel et al, 2010) considered the COM trajectory as a performance variable but only at heel strike. Three studies (Krishnan et al, 2013;Robert et al, 2009;Rosenblatt et al, 2014) analysed the temporal evolution of the UCM approach through the gait cycle but none did this with the COM movement as the performance variable.…”

Discrimination of gender-, speed-, and shoe-dependent movement patterns in runners using full-body kinematics

“…Different sets of kinematic data can be used to generate an input matrix. Recently, position data for limbs [9,10,12,13] or joint angles have been used [14][15][16][17]. Principal component analysis can be applied per condition [12][13][14]16], per subject [9,14,16,17], or across subjects [10,12,[15][16][17].…”

“…Principal component analysis can be applied per condition [12][13][14]16], per subject [9,14,16,17], or across subjects [10,12,[15][16][17]. In interpreting the data, one can use the variability explained by individual PMs [12][13][14][15][16][17]. It is possible to examine the change in PM for different conditions [9,10,[12][13][14]17], or to interpret the change in PVs for different conditions [9,10,15,16,18].…”

“…In interpreting the data, one can use the variability explained by individual PMs [12][13][14][15][16][17]. It is possible to examine the change in PM for different conditions [9,10,[12][13][14]17], or to interpret the change in PVs for different conditions [9,10,15,16,18]. The PMs and PVs for different conditions can be differentiated using classification methods such as support vector machine (SVM) or spherical classification [19,20].…”

“…In using this approach with data from running, the first five PMs often explain more than 95% of the variability and, therefore, represent the basic movements of the task [9,10,12,13]. Higherorder PMs are often neglected, as they explain only a small fraction of the variability and are considered to represent random noise.…”