Adaptive Autoregressive Modeling Based Ecg Signal Analysis for Health Monitoring
Varun Gupta, Monika Mittal, Vikas Mittal, Anshu Gupta
Abstract
In the present era, efficient analysis of electrocardiogram (ECG) signal is still a challenge due to large variations in its morphology. Therefore, it requires the proper utilization of digital signal processing (DSP) techniques to analyze raw ECG signals. Presently, cardiologists need the active involvement of computers equipped with efficient DSP techniques 2like pre-processing, feature extraction, and classification. In this chapter, pre-processing is performed using wavelet transform (WT) due to its better time-frequency resolution. Adaptive autoregressive modeling (AARM) is used for extracting features as its parameters are allowed to vary in time. And support vector machine (SVM) technique is considered for classification due to its high modeling stability even for non-linear data. The performance of the proposed technique is evaluated on the basis of parameters such as sensitivity (Se), positive predictivity (Pp), accuracy (Acc), and detection rate (Dr). The proposed technique has secured Se of 99.95%, Pp of 99.95%, Acc of 99.93%, and Dr of 99.95%. The authors expect that the proposed technique may be successful in classifying all major types of arrhythmias that were not classified by the existing methodologies single-handedly.