Chaotic approach based feature extraction to implement in gait analysis

Chaotic systems, which are fundamentally described as the systems having sensibility to the initial conditions, are extensively used in a broad area ranging from economics to engineering sciences, from biomedical to space researches. There exist methods to analyze chaos and gives an estimation, which are very helpful for dynamic analyses of systems being nonlinear or cannot be modeled mathematically based on differential equations. This study focuses on extracting features based on nonlinear time series to describe problems leading to balance disorder. For this purpose, required data is collected from pressure sensors placed at certain locations under the sole. A two-step classification process is projected, where the first step will determine the subject to belong either to ‘healthy’ group or to ‘ill’. In case the latter is assigned, a second step will follow where we search for the ability of chaotic approach to provide features to define the specific illness. During the study, firstly the delay coordinates have been constructed form the obtained time series, hence the nonlinear time series were hardly difficult to construct a suitable mathematical modelling [1]. Thereafter, optimal time delay and embedding dimensions [2–4] have been chosen, and the delay coordinates’ analysis for the shape of attractors [3], Poincare Mapping [3], and Fractal Analyses [3, 5] have been examined in order to detect a feature to distinguish the type of illness.