Features Extraction from ECG Signals Using Unbiased FIR Filtering

An electrocardiogram (ECG) is a record of heart wave patterns, which is used to identify abnormalities related to the electrical activities of the heart. In this paper, we discuss applications of the unbiased finite impulse response (UFIR) estimation approach to denoise the ECG measurements and extract useful features from. One of the standard methods providing denoising and extracting features of the ECG signals is based on the one-step prediction. We show that better accuracy can be obtained by using the UFIR filter and the UFIR smoothing filter with adaptive averaging horizon length. To study the trade-off between the predictive, filtering, and smoothing UFIR filtering algorithms, we employ the ECG data related to normal heart beating.