Predicting the termination of the atrial fibrillation based on the nonlinear features of RR intervals

SUN Rong-Rong,WANG Yuan-Yuan

DOI: https://doi.org/10.3969/j.issn.0258-8021.2008.02.010

2008-01-01

Abstract:It may lead to not only the better understanding of the termination mechanism of arrhythmia but also the improved treatment of the sustained atrial fibrillation(AF) by recognizing and characterizing whether the AF is likely to terminate spontaneously or sustained.In this paper,RR intervals of the AF were characterized with two kinds of nonlinear analysis methods to predict the termination of the AF.One feature was the Lempel-Ziv complexity calculated from RR intervals.Other two came from the symbolic dynamics.Firstly RR intervals were transformed to symbol strings.Then symbol codes were obtained by coding symbol strings.After the occurrence probability of all possible symbol codes were calculated,the occurrence probability of the symbol code 110 and the Shannon entropy of the occurrence probability of symbol codes were taken as two parameters of RR intervals.Based on the above three nonlinear parameters which were extracted from RR intervals,the fuzzy classifier was set up to predict the termination of the AF.An AF database which includes a training set and two testing sets(A and B) of Holter Electrocardiograph(ECG) recordings was used to verify the proposed method.Results showed that 90% of the testing set A and 80% of the testing set B were correctly classified.Compared to previous methods,the proposed method improved the prediction accuracy for the AF termination by approximately 10%.It is demonstrated that the algorithm has the ability to predict spontaneous termination of the AF from Holter ECG recordings effectively.