Simple wavelet-based features for arrhythmia identification from HRV signals based on an artificial immune system
Caio Cesar Enside de Abreu1; Bruno Rodrigues de Oliveira2; Marco Aparecido Queiroz Duarte3; Jozue Vieira Filho2; Francisco Villarreal2
1 State University of Mato Grosso; 2 São Paulo State University; 3 State University of Mato Grosso do Sul
Resumo
This paper presents two simple wavelet-based features for automatic arrhythmia identification from heart rate variability (HRV) signals. Decomposition provided by the stationary wavelet transform is used in order to filter possible changes in the HRV signal baseline and highlight abrupt fluctuations between two consecutive RR-intervals. From detail wavelet coefficients, two simple features named Range and Peaks are derived and used as input to a classifier. The classifier implemented is an artificial immune system (AIS) that has not yet been applied to the related problem. Results showed that proposed features, as well as the AIS are suitable for arrhythmia identification from HRV series, pointing that future research should be conduced based on new combinations of features encompassing the proposed ones.