Heartbeat recognition using extreme learning machine

The electrocardiogram plays a significant part in defining irregular activity in the patient's heart and is applied in the realization of heart disease. This ECG can be categorized as normal and abnormal signals. In this study, the ECG signals classification will be performed with Extreme Learning Machine, we extracted the feature vector for every beat from the dataset of ECG as the input of the ELM model. We will add white Gaussian noise to the ECG signals and de-noising the white Gaussian noise from the ECG signal by using such a wavelet filter, and then we compared the accuracy with other methods. This experiment was conducted on the ECG data to test a set of 8 normal ECG records and 26 abnormal ECG records, the database has been used our ELM-based classifier for training and testing. Besides, the method suggested achieved a satisfactory degree of precision in classifying ECG pulse and can be utilized in cardiology programs for cardiologists. Our algorithm simulation results demonstrate 90.9091% accuracy. Key Words: Extreme Learning Machine, Heartbeat Classification, Feature, Electrocardiogram (ECG) Signals Classification, White Gaussian Noise, Wavelet Filter