Improved ECG heartbeat classification based on 1-D convolutional neural networks

ECG (Electrocardiogram) waves have a significant role in identifying the nature of heart diseases and monitoring the heart situation of patients who suffer from different cardiovascular diseases. In this research, a model that implements a binary classification in the light of the 1D-CNN algorithm is proposed for the ECG signal environment. The main objective of this study is to detect and separate regular and irregular heartbeat signals. The model has been trained and tested on the MIT-BIH dataset and classifies the signal into normal and abnormal samples using various activation functions for different epochs. Test results indicated that, according to the confusion matrix, an improvement of 99.8%, 99.9%, and 99.7% was achieved for accuracy, sensitivity, and specificity, respectively. These results are better than other related works that have been recently published for the classification of ECG signals.

Automated summary

A model based on the 1D-CNN algorithm is proposed for the binary classification of ECG signals to detect and separate regular and irregular heartbeat signals. Test results on the MIT-BIH dataset showed significant improvements in accuracy, sensitivity, and specificity compared to other related works.