Recurrent neural networks with composite features for detection of electrocardiographic changes in partial epileptic patients.

The aim of this study is to evaluate the diagnostic accuracy of the recurrent neural networks (RNNs) with composite features (wavelet coefficients and Lyapunov exponents) on the electrocardiogram (ECG) signals. Two types of ECG beats (normal and partial epilepsy) were obtained from the MIT-BIH database. The multilayer perceptron neural networks (MLPNNs) were also tested and benchmarked for their performance on the classification of the ECG signals. Decision making was performed in two stages: computing composite features which were then input into the classifiers and classification using the classifiers trained with the Levenberg-Marquardt algorithm. The research demonstrated that the wavelet coefficients and the Lyapunov exponents are the features which well represent the ECG signals and the RNN trained on these features achieved high classification accuracies.