KPCA and SVR-based Cardiac Arrhythmia Classification on Electrocardiography Waves

Patients with numerous cardiac conditions may pass away suddenly. Therefore, early arrhythmia diagnosis is crucial to preventing unexpected fatalities. It is necessary to develop techniques for extracting ECG signal characteristics to diagnose cardiac irregularities and other illnesses. This paper demonstrates the kernel-PCA and supports the vector regression-based arrhythmia detection method. The T, S, Q, R, P beats, and RR intervals with a certain amplitude are among the features of an ECG signal. Several techniques have been put forth to identify and evaluate ECG signals. This study combines kernel-PCA with a support vector regression classifier to detect the ECG signal. The process has three phases: initially, a low-pass filter reduces noise from the ECG data. Later, for feature extraction of the ECG signal, higherorder statistics and Kernel-PCA are produced. The ECG wave is finally classified using the resulting feature set as input to the SVR classifier. Most of the information used to assess the classification performance comes from the online MIT-BIH dataset where ECG recordings were obtained. The suggested model has better classification in categorization performance when compared to several traditional machine learning classifiers.