Analysis and Prediction of Cardiovascular Disease Using Machine Learning Techniques

Abstract

Cardiovascular disease is the silent killer of human lives in the world. This disease has some common treatments such as meditation and health valve surgery. Prior to the treatment of the disease, we must know the patients have the symptoms or not. In the hospital system, a lot of patient records are available to analyze using machine learning algorithms. Our paper used heart disease patient dataset and six different kinds of supervised machine learning techniques that are k-NN, decision trees (DT), SVM, logistic regression, Naive Bayes (NB), and random forest to compare the accuracy, precision, recall, and f-measure on each classifier by train-test split as well as k-fold cross-validation methods with different ratio and values. Logistic regression model gives the best accuracy of 82% for 80:20 and 75:25 split. SVM also gives an accuracy of 82% for 75:25 split. Similarly, logistic regression model gives an accuracy of 82% for tenfold cross-validation, as the data is evenly distributed. In general, logistic regression and SVM have better accuracy than the other classifiers.