Automated ECG Delineation using Machine Learning Algorithms

The aim of automated electrocardiogram (ECG) delineation system is the reliable detection of fundamental ECG components and from these fundamental measurements, the parameters of diagnostic significance, namely, P-duration, PR-interval, QRS-duration, QTinterval, are to be identified and extracted. In this work, two supervised machine learning algorithms, K-Nearest neighbour (KNN) and Support Vector Machine (SVM) have been applied for accurate and efficient delineation of ECG signals. The algorithms were evaluated on a standard database CSE DS-3. The mean and standard deviations of the basic intervals obtained by KNN and SVM algorithms have been calculated and compared with three 12-lead programs used in the CSE study from the combined program median. The results show that the proposed algorithms give a new direction of using KNN and SVM effectively for the identification and delineation of the ECG wave components.