Hybradization of Emperical Mode Decomposition and Machine Learning for Categorization of Cardiac Diseases

The arrhythmia is one of the cardiovascular diseases which has several types. In literature, researchers have presented a broad study on the strategies utilized for Electrocardiogram (ECG) signal investigation. Automated arrhythmia detection by analyzing the ECG data is reported using a number of intriguing techniques and discoveries. In order to effectively categorize arrhythmia, a novel approach based on the hybridization of the denoising filter, QRS complex segmentation, « Empirical Mode decomposition » (EMD), « Intrinsic Mode Functions » (IMFs) based features extraction, and machine learning techniques is developed in this study. To evaluate the categorization accuracy, the 10-fold cross validation (10-CV) strategy is used. Using an arrhythmia dataset that is publically available for research, the performance of our method is evaluated. A 97% average accuracy score is secured by our method for the problem of 5-class arrhythmias. These findings are comparable or better than counterparts.