Traditional analysis of running gait utilizes averaged biomechanical data from several strides to generate a mean curve. This curve is then used to define the average picture of a runners gait. However, such measures are frequently accompanied by time normalization, which results in a loss of temporal variations in the gait patterns. An examination of stability requires analysis of both time and position, therefore loss of such information makes stability analysis difficult. On the contrary, the use of a dynamical systems approach for gait analysis allows for a better understanding of how variations in gait pattern change over time. In the current study runners ran on a treadmill, with both a flat and uneven surface, at a self selected speed. Three-dimensional position data was captured for 11 different anatomical locations at a frequency of 120 Hz using a Qualysis motion capture system. The data was first shifted to a lumbar coordinate system to account for low frequency drift attributed to the subjects’ drift on the treadmill. Since all of the markers were rigidly connected, via the subject, the movements and variations of certain components of the 33-dimensional measurements were not independent. As a result, it was possible to reduce the dimensionality of the transformed data using singular value decomposition techniques. The primary components were then analyzed using the method of delay embeddings to extract geometric information, revealing the natural structure found in the data as a result of the periodicity of each running stride. A nearest neighbor mean stride orbit was then computed to create a reference orbit, so that deviations from the mean stride orbit can be measured. The expectation was that a more stable running configuration would lead to smaller deviations from the mean stride orbit. On-going work that will be reported includes: (i) analysis of running stability related to the reference stride comparator, (ii) compensation of lumbar centroid dynamics, (iii) reconstructions using one dimension from the lumbar centroid transformed data, and (iv) consideration of transients, fatigue, adaptation, etc.

This content is only available via PDF.
You do not currently have access to this content.