SVD and SVM based approach for congestive heart failure detection from ECG signal

A detection approach integrating R-R interval features extraction and classification for congestive heart failure (CHF) is presented in this paper. In R-R interval features extraction, we use empirical mode decomposition (EMD) to decompose each subject's R-R interval signal into several intrinsic mode functions (IMF), and use singular value decomposition (SVD) to extract the ranked singular values for each subject's IMF. The ranked singular values obtained are input to the support vector machine (SVM) for classification of physiological state (health or CHF). The classification results showed that, based on a dataset from the “PhysioNet” website, the total accuracy of using significant ranks' singular values as features was 89%. These results indicate a promising method of combining SVD and SVM in distinguishing healthy persons from CHF patients.