Recurrent neural networks employing Lyapunov exponents for analysis of ECG signals

An approach based on the consideration that electrocardiogram (ECG) signals are chaotic signals was presented for automated diagnosis of electrocardiographic changes. This consideration was tested successfully using the nonlinear dynamics tools, like the computation of Lyapunov exponents. Recurrent neural network (RNN) was implemented and used as basis for detection of variabilities of ECG signals. Four types of ECG beats (normal beat, congestive heart failure beat, ventricular tachyarrhythmia beat, atrial fibrillation beat) obtained from the PhysioBank database were classified. Decision making was performed in two stages: computing features which were then input into the RNN and classification using the RNN trained with the Levenberg–Marquardt algorithm. The research demonstrated that the Lyapunov exponents are the features which are well representing the ECG signals and the RNN trained on these features achieved high classification accuracies.