Heart Disease Prediction under Machine Learning and Association Rules under Neutrosophic Environment

Abstract

Early identification and precise prediction of heart disease have important implications for preventative measures and better patient outcomes since cardiovascular disease is a leading cause of death globally. By analyzing massive amounts of data and seeing patterns that might aid in risk stratification and individualized treatment planning, machine learning algorithms have emerged as valuable tools for heart disease prediction. Predictive modeling is considered for many forms of heart illness, such as coronary artery disease, myocardial infarction, heart failure, arrhythmias, and valvar heart disease. Resource allocation, preventative care planning, workflow optimization, patient involvement, quality improvement, risk-based contracting, and research progress are all discussed as management implications of heart disease prediction. The effective application of machine learning-based cardiac disease prediction models requires collaboration between healthcare organizations, providers, and data scientists. This paper used three tools such as the neutrosophic analytical hierarchy process (AHP) as a feature selection, association rules, and machine learning models to predict heart disease. The neutrosophic AHP method is used to compute the weights of features and select the highest features. The association rules are used to give rules between values in all datasets. Then, we used the neutrosophic AHP as feature selection to select the best feature to input in machine learning models. We used nine machine learning models to predict heart disease. We obtained the random forest (RF) and decision tree (DT) have the highest accuracy with 100%, followed by Bagging, k-nearest neighbors (KNN), and gradient boosting have 99%, 98%, and 97%, then AdaBoosting has 89%, then logistic regression and Naïve Bayes have 84%, then the least accuracy is support vector machine (SVM) has 68%.