Classification of cardiovascular diseases using pcg

Cardiovascular diseases have been reported to be the major cause of death world wide. According to the study conducted by Global Burden of Disease, fatality rate of 25% has been reported in India. This alarming statistic drives home the fact that one is more likely to die of a heart attack than a road accident and the need for a machine learning model that aids in the easy recognition of abnormalities in the heart. This model implemented in the present study has the following stages viz: Signal enhancement, Parameter extraction, Classification, and Evaluation. The total 33 different parameters from time domain, frequency domain and statistical parameters are considered in this work. Three different classifiers viz K-Nearest Neighbours, Ensemble and Support Vector Machine (SVM) have been explored. This paper analyzes performances of these machine learning models and validates them using the available phonocardiogram (PCG) signal data set. Hence this study propose an efficient method for categorizing normal against abnormal heart sound recordings from the given PCG signals.