A Review of Role of Machine Learning Models in Coronary Heart Disease Detection Accuracy

Abstract

According to the World Health Organization, heart disease is the most widespread disease in the world, affecting over a billion people. Generally, the lifestyles of people are occasionally plagued by stress, worry, and sadness, among other things. The early detection of this condition is tough, and it is a difficult task in medical science. The goal of this research is to better understand the detection accuracy of particular machine learning models (MLMs), as well as their limitations and categorization strategies. Many researchers used classification techniques such as Naive Bays (NB), decision trees (DT), Cooperative Neural-Network Ensembles (CNNEs), logistic regression (LR), Support Vector Machine (SVM), Least Square Twin Support Vector Machine (LS-SVM), k-Nearest Neighbor (KNN), Bays Net (BN), Artificial Neural Network (ANN), and Multi-Layer Perception (MLP) (MLP). In total, the dataset contains more than 50 features attributes. To boost accuracy, the study uses different feature selection approaches to identify the most appropriate features for detecting the disease. The present study achieved a maximum classification accuracy of 96.29%, and there is a need to improve accuracy in the shortest period possible by developing single MLMs for detecting and selecting specific features. Many studies employ hybrid approaches to improve the accuracy of percentages by layering two or more classification algorithms (based on specified symptoms and traits of a human being). It is not always more efficient and time-consuming. As a result, flexible MLMs with feature selection and reduction strategies are required. Further, the current research focuses on boosting accuracy and includes future viewpoints or uses of research as well.