ABSTRACT
In a world of increasing prevalence of Cardio Vascular Disease (CVDs) and alarmingly high mortality worldwide, it is therefore important to establish early detection strategies. Phonocardiogram (PCG) indications have historically been the mainstay of the industry, largely because they were inexpensive and easy to use. In this study, we present PhysioNet - a novel CNN and BiLSTM method designed to automate the classification of different sounds of cardiac auscultations such as normal, aortic stenosis (AS), mitral stenosis (MS), mitral regurgitation (MR) and mitral valve prolapse (MVP) with a high accuracy. We automated the technique by applying those two learning phases : Representation Learning & Sequence Residual Learning. This projects main aim to classify cardiac disease with PCG (Phonocardiogram) data. The purpose of this work is predicting the four pathological classes in cardiovascular disease via auscultatory sounds and deep learning technology. So finally, you will predict the Disease class based on that variable. Calculate the accuracy, precision, recall and F1 score for each of those predicted classes over the corresponding WAV file.
