Classification Heart Diseases Base on Heart Sound Using Backpropagation Algorithm

Abstract

Heart sounds have a special pattern that can indicate a person's heart condition. An abnormal heart will produce a characteristic sound that is often called a murmur ( Sitinjak , 2008). Murmurs are caused by various things that can indicate a person's heart condition. From these murmurs can be known the types of abnormalities experienced by patients. In this study, cardiac abnormalities that can be identified are aortic stenosis (as), mitral regurgitation (mr), mitral valve prolapse (mvp), mitral stenosis (ms), and normal. The data used for training as many as 1000 heart sound files consisting of 200 files each for each heart abnormality.Data in the form of heart rate sound samples with the format. Wav. The program was created using the Artificial Neural Network method to identify the five types of cardiac abnormalities. The training method is created using the traingdx function provided in the Neural Network Toolbox on MATLAB. Based on the results of the training can be obtained a validity value of 97,7%.