Prediction and classification of cardiac arrhythmia using artificial intelligence based on complex wavelet electrocardiographic extraction

Abstract

Cardiovascular diseases contribute to the major cause of the death rate in modern days. These diseases do not have any specific symptom often. They may cause shortness of breath or chest pain. But it can be predicted at an early stage so that by providing proper medication, the death due to heart diseases may be reduced. The heartbeat follows a rhythm. The disruption in the normal heartbeat rhythm is called Cardiac Arrhythmia. Most of the heart diseases are indicated by the change in the pattern of the heartbeat. If the heartbeat is abnormally slower, then it is called bradycardia, and if it beats abnormally faster, then it is called tachycardia. The process of recording the electrical activity of the heart is known as Electrocardiography (ECG). The ECG uses 12 electrodes placed on the surface of the chest and limbs of the patient. Sometimes, it may be measured by using three electrodes based on the principle of Einthovenand#8223;s triangle. The ECG signals are affected by the external noises like power line interference, baseline wandering, muscular artifacts and electrode artifacts. These noises may leads to misclassification and provide false results. Hence it is important to remove these artifacts from the ECG signals to have a proper prediction of cardiac diseases. The ECG signals are taken from the MIT-BIH database is corrupted with noises. To remove these noises, preprocessing is required. Preprocessing uses filters which will eliminate the high-frequency signals and allows only the low-frequency heart signals for further processing The proposed Discrete Orthonormal Stockwell Transform (DOST) with fuzzy kNN and kNN classifier involves the extraction of the features from the ECG signal and classification of various types of arrhythmias in the ECG signals. newline