Intra-session reliability of local dynamic stability of walking

Abstract

While local dynamic stability measures have been successfully used to characterize walking stability, they require long continuous walking data, which may be difficult to obtain from a clinical population. We investigated the amount of walking data necessary to obtain reliable measures of local dynamic stability. Twenty healthy adults walked on a motorized treadmill at their self-selected speed for three trials of 5 min each. Trunk motion was used to construct a 12-dimensional state space comprised of the linear and angular positions and velocities. Mean divergence of locally perturbed trajectories was calculated as a measure of local dynamic stability using the first 1-5 min of data from each trial. Exponential divergence rates were quantified. Divergence was also parameterized using a double-exponential function. Intra-class correlation coefficients ICC(2,1) were calculated for each divergence measure for each trial length. ICC(2, 1) values increased with trial length, and reached 0.5-0.9. Good reliability was obtained for short-term measures for trial lengths of 2 and 3 min, but 5 min was not adequate to estimate the long-term coefficients based on a single trial.