Morphological heart arrhythmia detection using Hermitian basis functions and kNN classifier

This paper presents the results of morphological heart arrhythmia detection based on features of electrocardiography, ECG, signal. These signals are obtained from MIT/BIH arrhythmia database. The ECG beats were first modeled using Hermitian basis functions, (HBF). In this step, the width parameter, sigma, of HBF was optimized to minimize the model error. Then, the feature vector which consists of the parameters of the model is used as an input to k-nearest neighbor, kNN, classifier to examine the efficiency of the model. In our experiments, seven different types of arrhythmias have been considered. We achieved the sensitivity of 99.00% and specificity of 99.84% which are comparable to previous works. These results were obtained in less than 0.6 second which is suitable for real-time diagnosis of heart arrhythmias.