Utilizing occurrence sequence of Heart Rate's phase space points in order to discriminate heart Arrhythmia

Abstract

Poincare plot analysis of RR time series allows a beat-to-beat approach to Heart Rate Variability (HRV), detecting patterns associated with nonlinear processes. Since the measurement of standards descriptors of Poincare plot is based on the point's distribution in relation to the line of identity (y=x), we have concentrated on it and evaluated the points behavior related to this line. For this purpose, we test two global and local analyses of points against the identity line. For evaluating these two novel features of Poincare plot, we try to use them for distinguishing four groups of subjects (Arrhythmia, Congestive Heart Failure (CHF), Atrial Fibrillation (AF) and Normal Sinus Rhythm (NSR)). Kruskal-Wallis test was used to define the level of significance of features. The results show that global feature discriminate different groups by p<6E-7, and local feature discriminate them by p<2E-7.