Dual State-Parameter Estimation of ECG Signals with Recursive Bayesian Filters

Abstract

The Electrocardiogram (ECG) is used for diagnosing heart conditions by recording the small electric waves generated during heart activity. Gois et al. have recently proposed a mathematical model to describe heart rhythms considering three-coupled Van der Pol oscillators, and indicated that the heart rhythms of the cardiac diseases can be shown by changing three coupling parameters. In this paper, we propose two methods for simultaneous estimation of the states and the parameters of Gois ECG model. For the joint state-parameter estimation, we constitute the ensemble Kalman filter from this model to detect the cardiac diseases by estimating the coupling parameters as the state variables in Gois’s model. For the dual space-parameter estimation, in addition, we apply the dual ensemble Kalman filter to estimate the coupling parameters. Finally, we compare the estimation results of these filters.