Diagnosis of Coronary Artery Disease Using Data Mining Based on Lab Data and Echo Features

According to American Heart Association report, cardiovascular diseases are one of the five leading causes of death in the world. Coronary Artery Disease (CAD) is the most common fatal heart disease, and is the subject of large body of studies. According to prevalence of CAD, early diagnosis of this disease is very important. The most reliable method for CAD diagnosis is angiography, but it is costly, time-consuming, and hazardous. Therefore in order to predict such diseases, study of non-invasive methods such as analysis and mining of patients' medical information is becoming popular, and has proved to be effective. Unfortunately, majority of approaches in the literature rely on a limited and small set of medical features for disease detection. This paper aims to examine effects of set of features; including lab data and echo information on CAD diagnosis which some of them were not considered in previous studies. The data set consists of the information gathered from 303 random visitors to Tehran's Shaheed Rajaei Cardiovascular, Medical and Research Center which is one of the largest heart hospitals in Asia. The method used in this research was data mining. Several classification algorithms were adopted to analyze the data set, including SMO, Naive Bayes, C4.5 and AdaBoost. According to the comprehensive set of features used, the obtained classification accuracy exceeded 82 percent. Results showed that new added features including Region with RWMA and Ejection Fraction (EF) have a large effect on CAD.