A study on predicting and diagnosing non-communicable diseases: case of cardiovascular diseases

Abstract

Heart disease causes millions of deaths worldwide. Many approaches have been proposed for the prediction of heart disease. Several machine learning, deep learning, and data mining algorithms are used in the detection and diagnosis of heart disease based on parameters or risk factors. The most used algorithms are Naïve Bayes, Machine Vector Support, decision tree, KNNs, and artificial neural networks. The most frequently used parameters or risk factors are the 14 attributes of the UCI Cleveland standard. In this article, a study on these different approaches is carried out. This study shows diversity in relation to the choices and the use of different attributes in the prediction of cardiovascular diseases.