Does human gait exhibit comparable and reproducible long-range autocorrelations on level ground and on treadmill?

Stride duration of young healthy subjects walking at spontaneous speed fluctuates over the long-term in a very complex way. The presence of long-range autocorrelations among these fluctuations has already been highly suggested for subjects walking on level ground, but the mathematical methods used among studies are variable. Moreover, despite the frequent use of a treadmill in research and in rehabilitation, the presence of such autocorrelations was nearly exclusively assessed during level-ground walking. The first aim of this study was to confirm the presence of long-range autocorrelations among stride duration variability on level ground with a good level of confidence. The stride duration fluctuations of 10 young healthy subjects were assessed on a 37-meter-long track using an integrated approach that combines the results of rescaled range analysis and power spectral analysis. Secondly, the results obtained from treadmill tests were compared with the outcomes of the level-ground walking tests in order to challenge the persistence of the long-range autocorrelations during treadmill walking. Finally, the third aim of this study was to test the reproducibility of the outcomes by comparing the results obtained on the treadmill during two different trials. The outcomes of this study provide significant evidence to confirm the hypothesis that long-range autocorrelations are present among stride duration variability, not only on level ground but also on the treadmill. The reproducibility of the results obtained during treadmill walking further validates the use of a treadmill to assess the long-term fluctuations of gait.