Effective ECG beat classification using higher order statistic features and genetic feature selection

One of the most significant indicators of heart disease is arrhythmia. Detection of arrhythmias plays an important role in the prediction of possible cardiac failure. This study aimed to find an efficient machine-learning method for arrhythmia classification by applying feature extraction, dimension reduction and classification techniques. The arrhythmia classification model evaluation was achieved in a three-step process. In the first step, the statistical and temporal features for one heartbeat were calculated. In the second, Genetic Algorithms (GAs), Independent Component Analysis (ICA) and Principal Component Analysis (PCA) were used for feature size reduction. In the last step, Decision Tree (DT), Support Vector Machine (SVM), Neural Network (NN) and K-Nearest Neighbour (K-NN) classification methods were employed for classification. The proposed classification scheme categorizes nine types of Electrocardiogram (ECG) beats. The experimental results were compared in terms of sensitivity, specificity and accuracy performance metrics. The K-NN classifier attained classification accuracy rates of 98.86% and 99.11% using PCA and ICA features. The SVM classifier achieved its best classification accuracy rate of 98.92% using statistical and temporal features. The K-NN classifier feeding genetic algorithm features achieved the highest classification accuracy, sensitivity, and specificity rates of 99.30%, 98.84% and 98.40%, respectively. The results demonstrated that the proposed approach had the ability to distinguish ECG arrhythmias with acceptable classification accuracy. Furthermore, the proposed approach can be used to support the cardiologist in the detection of cardiac disorders.