A Comparative Case Study of Balance Dynamics in Parkinsonian Patients
Abstract: This research work explores several methodologies to quantify differences in balance dynamics between Parkinsonian and non-Parkinsonian (control) subjects. These methodologies are based on, respectively, correlational, spectral, and complexity analyses, and take advantage of the time-dependent data obtained by acceleration measurements for the subjects asked to maintain standing postural balance on different surfaces. In this work we provide a computational quantitative assessment of the complexity of random time series via the tool called \epsilon-complexity. We have used binary (Parkinson's Disease or Control) classification procedure based on the data set consisting of several estimated parameters obtained by the complexity analysis of the acceleration data for all subjects. Although all these methodologies provide clear separation between the Parkinsonian subjects and controls, it is the complexity analysis of the acceleration signals that separates the two categories most efficiently.