Detection of Coronary Artery Disease Using C4.5 Decision Tree

Today, one of the most common diseases and causes of death in the world is heart diseases. Data mining techniques are very useful to create predictive models for identifying people at risk and decreasing the disease complications. In this study, using C4.5 decision tree method, the prevention and diagnosis of this disease are discussed.
This was an applied descriptive study. UCI standard data and Cleveland data collection were used. The database contains 297 records. Analysis was performed through Weka software and using CRISP3 methodology. The C4.5 decision tree model, using input variables and determining the target variable, was created.
According to the applied model, it was found that high levels of cholesterol, sex, age, high maximum heart rate, scan thallium higher than 3 and abnormal ECG have the greatest impact on the risk of coronary heart disease. Furthermore, by using the created decision tree, some rules were extracted that can be used as a model to predict the risk of coronary heart disease. The accuracy of the model created by using decision tree was over 80 percent.
According to our calculations, the rate of categorization was 72.6% and the accuracy of C4.5 algorithm was 80.2% that in comparison with the results of studies in the field of data mining of heart diseases, the obtained accuracy for the suggested algorithm is acceptable.