Figure 14

Surrogation analysis. Surrogate data for step width and step length time series for each subject (#1–10) using the fractal scaling exponent (α-DFA) obtained from detrended fluctuation analysis (DFA) as the discriminating statistic. Ten-minute step length time series before (a) and after random permutation (b), and detrended fluctuation analyses (c) are shown for a representative walking trial with supra-threshold transspinal stimulation delivered at 15 Hz. The solid-colored lines in (c) are the best fit used to calculate the α-DFA value of the original and permuted time series. For the surrogate data test, the step width and step length time series of each walking trial was randomly permutated 399 times and the α-DFA value was calculated each time. The distribution of the surrogate α-DFA values (399 data) for each walking trial is shown in (d) for the step width and in (e) for the step length time series. (f) Step length and step width without transspinal stimulation for each subject. The α-DFA value of the original time series is shown with a vertical line, along with the probability density functions and the quantiles (2.5% and 97.5%) of the surrogates. Asterisks indicate that the α-DFA value for the original time series was significantly greater than α-DFA values for the corresponding surrogate time series, indicating a significant difference from uncorrelated white noise (*p ≤ 0.005).