Real-time Heart Disease Prediction System using Multilayer Perceptron

Stroke, Heart Failure, Arrhythmia, and myocardial infarction are the most common cardiovascular diseases which record high mortality rates around the world. Heart defects are not detected in the early stages due to the impractical costs of the tests available. Thus, a fast, real-time, and reliable system that predicts the chances of a patient having heart disease in an optimized manner is required.In this research, a Neural Networks model using a Multilayer Perceptron (MLP) is proposed for the prediction system. Experimental analysis resulted in an accuracy of 85.71% for UCI Heart Disease dataset and 87.30% for Cardiovascular Disease dataset. When compared to previous research the increase in accuracy was approximately 12-13%. A simple web application tool is also developed using python programming to test the prediction system. This research works towards making a comprehensible tool for medical professionals as well as common people.