Benefits of nonlinear analysis indices of walking stride interval in the evaluation of neurodegenerative diseases.

Abstract

Indices characterising the long-range temporal structure of walking stride interval (SI) variability such as Hurst exponent (H) and fractal dimension (D) may be used in addition to indices measuring the amount of variability like the coefficient of variation (CV). We assess the added value of the former indices in a clinical neurological context. Our aim is to demonstrate that they provide a clinical significance in aging and in frequent neurodegenerative diseases such as Parkinson's disease, Huntington, and amyotrophic lateral sclerosis. Indices assessing the temporal structure of variability are mainly dependent on SI time series length and algorithms used, making quantitative comparisons between different studies difficult or even impossible. Here, we recompute these indices from available SI time series, either from our lab or from online databases. More precisely, we recompute CV, H, and D in a unified way. The average SI is also added to the measured parameters. We confirm that variability indices are relevant indicators of aging process and neurodegenerative diseases. While CV is sensitive to aging process and pathology, it does not discriminate between specific neurodegenerative diseases. H, which measures predictability of SI, significantly decreases with age but increases in patients suffering from amyotrophic lateral sclerosis. D, catching complexity of SI, is correlated with total functional capacity in patients with Huntington's disease. We conclude that the computation of H complements the clinical diagnosis of walking in patients with neurodegenerative diseases and we recommend it as a relevant supplement to classical CV or averaged SI. Since H and D indices did not lead to the same observations, suggesting the multi-fractal nature of SI dynamics, we recommend to open clinical gait analysis to the evaluation of more parameters.