Evaluation of walking pattern using acceleration signal

We have attempted to evaluate the gait of patients with Parkinson disease by using the fractal dimension of waist acceleration. This study was performed on 12 patients with Parkinson disease, and with six normal elderly volunteers. The acceleration signal was measured when the subject walked along a corridor. A wavelet-based fractal analysis method was used to estimate the fractal dimension of acceleration. The variances of wavelet coefficients were calculated at each scale, From the slope of logarithmic plots of the scales 6 to 3, we were able to estimate the parameter H, which is related directly to the fractal dimension. In the anteroposterior and vertical directions, the fractal dimensions of acceleration signals in patients with Parkinson disease were significantly higher than those of the normal volunteers. Our results suggest that the acceleration signals during walking have a rough structure in patients with Parkinson disease, and a smooth structure in normal elderly people.