Improving the Effectiveness of Heart Disease Diagnosis with Machine Learning

Abstract

Despite technological and clinical improvements, heart disease remains one of the leading causes of death worldwide. A significant shift in the paradigm would be for medical teams to be able to accurately identify, at an early stage, whether a patient is at risk of developing or having heart disease, using data from their health records paired with Data Mining tools. As a result, the goal of this research is to determine whether a patient has a cardiac condition by using Data Mining methods and patient information to aid in the construction of a Clinical Decision Support System. With this purpose, we use the CRISP-DM technique to try to forecast the occurrence of cardiac disorders. The greatest results were obtained utilizing the Random Forest technique and the Percentage Split sampling method with a 66% training rate. Other approaches, such as Naïve Bayes, J48, and Sequential Minimal Optimization, also produced excellent results.