Arrhythmia Disease Detection Using Grey Wolf Optimized Deep Belief Network

Arrhythmia is a heart condition that results from irregular heart electrical activity. An electrocardiogram (ECG) is a tool medical professionals use to determine heart disorders or arrhythmias. To do this, a computer-assisted diagnosis system is presented in this work. For ECG-based arrhythmia classification, a GWO-based DBN model is proposed using Grey Wolf Optimization (GWO) for training the Deep Belief Network (DBN). Moreover, the pre-processing is done by the Daubechies wavelet filter. Subsequently, feature extraction is performed by extracting the Fractional Fourier Transform (FrFT) and Discrete Wavelet Transform (DWT) from the signal. The proposed GWO-based DBN utilizes the extracted features to differentiate and classify the signals into two categories: arrhythmia and normal heartbeat. The performance of GWO-based DBN is evaluated using accuracy, sensitivity, and specificity. Finally, the outcome shows that the GWO-based DBN model attained better accuracy, sensitivity, and specificity rates of 0.954, 0.978, and 0.938, respectively.