Maximum Lyapunov exponent revisited: Long-term attractor divergence of gait dynamics is highly sensitive to the noise structure of stride intervals

Fig 2. Divergence curves. A 5-dimensional attractor was built through the delay-embedding theorem from the trunk acceleration signal. Divergence between neighboring trajectories in the attractor is shown. For each noise type, 109 curves from 69 individuals were aggregated. Original stride intervals were changed to four different noise structures (hybrid signals). Xaxis = time i normalized by stride. Y-axis = the logarithm of the i Euclidian distance d downstream of the j pair of the nearest neighbors in the attractor, averaged over all the pairs: .