Spectral feature and optimization- based actor-critic neural network for arrhythmia classification using ECG signal

Abstract

ABSTRACTArrhythmia classification is an interesting research field that serves as the solution for most of the cardiac-related diseases. The patients with cardiac diseases are experiencing the greatest risk rate of death, and hence, there is a need to identify the presence of arrhythmia in patients to reduce the fatality rate. This paper proposes an arrhythmia classification method, which offers better classification accuracy and releases the time spend for classifying the patients. The proposed method of arrhythmia classification uses the Electrocardiography (ECG) signal to classify the patients with and without arrhythmia. Initially, the wave components are identified from the ECG signal and are subjected to the feature extraction. The spectral and statistical features are extracted from the wave components that yield the texture and the geometric nature of ECG such that classification of ECG becomes effective. The classification is carried out using the Actor-Critic (AC) Neural Network that is trained using the Proposed Taylor-Sine Cosine Algorithm (Taylor-SCA). The Proposed Taylor-SCA algorithm is the integration of Taylor series and SCA. The experimentation is performed using the MIT-BIH Arrhythmia Database, and the experimental results show that the proposed algorithm exhibits the maximum accuracy, sensitivity, and specificity of 0.9545, 0.77, and 0.9375, respectively.