Gait Variability Analysis in Neurodegenerative Diseases Using Nonlinear Dynamical Modelling

Neurodegenerative diseases (NDDs) including Parkinson's disease (PD), Amyotrophic Lateral Sclerosis (ALS), Huntington Disease (HD) disrupt the neuromuscular control system and becomes one of the most serious implications of the human gait disturbance. Therefore, the early detection and classification of such diseases is crucial which could change the course of the treatment. Therefore, this paper explores the improvement of the classification capability based on number of features extracted from vertical ground reaction force (VGRF) signal using a nonlinear dynamical signal analysis technique; reconstructed phase space and recurrence quantification analysis (RQA). To remove any correlation, features have been orthogonally transformed using principal component analysis (PCA) in order to improve the classification performance. Support vector machine (SVM) with radial basis kernel function (RBF) has been used in the classification process. Results show the robustness of the proposed techniques with an overall accuracy 92.19%.