Comparison of Machine Learning Models in Prediction of Cardiovascular Disease Using Health Record Data

Abstract

Cardiovascular diseases (CVDs) are disorders of heart and blood vessels. This is the leading cause of deaths worldwide. Early detection and diagnosis can help the patients. Machine learning can be used to create the predictive model using cardiovascular diseases risk factors like cholesterol level, glucose level and blood pressure. The aim of this study is to do comparison of machine learning models on prediction of cardiovascular disease using patients' cardiovascular risk factors. Source of data is kaggle machine learning competitions, the dataset consists of 70,000 patients records. The machine learning algorithms used in this study are Random Forest, Naive Bayes, KNN and Logistic Regression. The results of comparison shows that Random Forest achieve high classification accuracy of 73 %, specificity of 65% and sensitivity of 80%. The model can be used in medical field for prediction of cardiovascular diseases.