Hilbert transform and neural networks for identification and modeling of ECG complex

This paper presents a method for modeling and identification of electrocardiogram signals; the proposed method consists of two phases; the first one is focused on obtaining the period of an ECG signal using a procedure of autocorrelation. The second phase obtains R-peaks using the Hilbert transform. Finally, an Artificial Neural Network using a retraining technique is applied for the prediction stage; this has been validated using the record 100 from the MIT-BIH arrhythmia database. Results confirm that the presented approach for detection of the ECG complex obtains 100% accuracy. The performance of the prediction method is promising due to the root mean squared errors of the prediction are of 0.029, 0.04, and 0.059 of the ECG amplitude, for 1, 2, and 3 steps ahead, respectively.