Nonlinear time series analysis: Application to cardiac diseases

The analysis of ECG signals with methods derived from chaos theory is a potential toolto classify different heart behaviors and can help to get insights on the heart dynamics.The main purpose of the present work are to implement and validate the correlationdimension (D2) method for HRV analysis and to investigate whether it is possible todistinguish between the HRV-signals of healthy subjects and heart diseases subjects onlybased on the D(2) or whether the Largest Lyapunov Exponent (LLE) can be used for thisaim. We used D(2) and LLE methods from nonlinear time series analysis to characterizehuman ECG signals obtained from the commercially available MIT-BIH ECG arrhythmiadatabase. Three groups of ECG signals have been considered: the ECGs of Normalsubjects and ECGs of subjects with Atrial Fibrillation (AF) and with PrematureVentricular Contraction (PVC). The correlation dimension (D(2)) method is related tochaos theory and it used to quantify heart rate variability (HRV). The D(2) is a featuredescribing character of signals and often used for classification of signals (ECG).Lyapunov exponents measure the average local rate of divergence of neighboringtrajectories in phase space embedding and quantify the sensitivity of the system to initialconditions, which is an important feature of chaotic systems. A positive LyapunovExponents can be taken as a definition of chaos. D(2) and Largest Lyapunov exponent(LLE) are increasingly used to classify systems (say for diagnostics purposes). ECG timeseries were classified according to results obtained from computation of D(2) and LLE. Ourresults confirm the previous studies, which indicate that technique from nonlineardynamical systems theory should help us understand the mechanism underlying cardiacdiseases and allow one to distinguish between different groups of patients with moreconfidence than the standard methods for time series processing accepted in cardiology.Keywords: Nonlinear Dynamics, Time Series Analysis, Chaos, Correlation Dimension,Lyapunov Exponent, Heart Rate Variability.

Keywords:

Nonlinear Dynamics , Time Series Analysis , Chaos , Correlation Dimension , Lyapunov Exponent , Heart Rate Variability