Using K-nearest neighbor (KNN) machine learning algorithm to diagnose heart diseases in (Alkarama Teaching Hospital/ Iraq)

Abstract

Despite the presence of many research papers related to the prediction of heart disease through the use of artificial intelligence, the significance of heart disease and the extent of its impact on the lives of people lead to the fact that the researchers must continue to work on building the best program to predict heart disease from an early age to avoid its development, in this paper the researchers are using the K-nearest neighbor algorithm for the prediction of heart disease. The dataset used in this paper is collected from local hospitals in Iraq and was cleaned and analyzed by the researchers. The diagnosis system that is proposed in this work is to assist physicians to diagnose heart conditions by converting medical factors of the patients in to numerical representations, the simulation results show that the proposed K nearest neighbor classifier has 86% accuracy in classifying 4 medical heart conditions when using controlled databases. The researchers concluded that the most common heart conditions that are found in Iraq are the ones classified in this algorithm, which are Coronary heart disease, Congestive Heart failure, and Arrhythmias. The researchers conclude that there is a direct relationship between the amount of training data and the algorithm’s accuracy and an inverse relationship between the number of neighbors K and the algorithm’s accuracy.