ECG signal analysis by using Hidden Markov model

This paper explores the real-time normal and abnormal heartbeats recognition system mainly based on electrocardiogram (ECG). The recognition of heartbeats from electrocardiogram (ECG) is performed by a statis-tical model, Hidden Markov model (HMM), to immedi-ately determine the status of the patient´s heartbeats. The ECG features developed by existing papers are used to train the HMM model. The same features of testing data are then fed into the trained HMM model for recognition. The four abnormal heartbeats include the left bundle branch block (LBBB), the right bundle branch block (RBBB), the ventricular premature contractions (VPC), and the atrial premature contractions (APC) are recognized for the ECG data in the MIT-BIH Arrhythmia Da-tabase. Experimental results in this paper shown that the proposed system performed well and had very excellent recognition rate for some heartbeat cases.