SMART SOUND DETECTION SYSTEM FOR CLASSIFYING HEART DISEASE USING ARTIFICIAL NEURAL NETWORK

Abstract

Abstract
An early recognition device for heart abnormalities, for normal and murmur, is indispensable to help medical personnel diagnosing heart abnormalities. Every persons have a special heart sound characteristic, an abnormal heart sound called murmur. Murmur heart sound will generate a special sound, so using frequency and spectrum wave, that sound can be analyzed to be known as normal or murmur heart sound. This design intended to identify 2 kind of heart sound, that is normal and murmur. This system consist of hardware and software that created with several stages: identification of needs, analysis of needs, system design, steps of making, program flow diagram, testing and data retrieval. The hardware using acoustic stethoscope, voltage regulator module, cutoff frequency circuit and voice recording module that used to give heart sound input to be processed in system. The software using backpropagation from Artificial Neural Network method to identify normal or murmur heart sound. This training method using nprtool function from MATLAB neural network toolbox. Based on the obtained test result of this diagnose device, the training accuracy is already 100% in recognizing 2 kinds of heart sound, that is normal and murmur. After testing the device as whole, system circuit is able to function well, based on the voice recording hardware that able to record heart sound and send it to be processed further in the software. In software testing, the buttons are able to function well too, with able to record heart sound, to filter heart sound, to extract heart sound feature, and analyze it's, to get a decision, to be in normal or murmur heart sound category.
Keywords: Heart Sound, MATLAB, Murmur, Artificial Neural Network, Backpropagation