Arrhythmia classification based on improved monarch butterfly optimization algorithm

Abstract

Heart attack is a major cause of mortality all over the world also heart-related diseases have increased the expenditure of health care. Electrocardiogram (ECG) is one of the simple ways to diagnose heart diseases. Arrhythmia is characterised by abnormalities in the rhythmic pace of heart, which may occur on a random basis in a everyday life of people. It was necessary to use long-term ECG recording devices to capture these rare occurrences. Morphological feature and wavelet coefficients-based features obtained from the recorded ECG signals. The feature vector is optimized by an Improved Monarch Butterfly optimization (IMBO) algorithm to reduce the dimensionality. These optimized features are applied to convolution neural networks to classify signals. The experimental results of the proposed method give a 99.49% accuracy, 99.58% sensitivity, and 98.83% specificity which is comparative to previous methods, and found better classification accuracy which may help the physician for diagnosing an arrhythmia.