Heart Disease Prediction using Machine Learning

A complex disease, heart disease affects many people worldwide. Fast and accurate cardiac disease detection is crucial in cardiology. This article proposes a machine learning-based cardiac disease diagnosis method that is efficient and accurate. The system utilizes classification algorithms like SVM, LR, ANN, K-nearest neighbor, NB, and Decision tree, as well as feature selection algorithms like Relief, MRMR, LASO, and Local learning to eliminate irrelevant and redundant features. We also suggested a fast conditional mutual information feature selection approach for feature selection. The features selection techniques improve classification accuracy and reduce classification system execution time. Additionally, leave one subject out cross-validation has been utilized to learn model evaluation best practices and hyper parameter tuning. Classifier performance is measured using metrics. Classifier performance was tested on features selected by features selection methods. Experimental results suggest that the proposed feature selection method (FCMIM) with classifier support vector machine can create a high-level intelligent heart disease detection system. More accurate than prior methods, the suggested diagnosis system (FCMIM-SVM) The proposed approach can also be readily deployed in healthcare to detect cardiac problems.