Revolutionizing Heart Disease Diagnosis with Superior Feature Selection: The Power of mRMR

Abstract

A serious concern to human and their health upcoming years is heart disease.  To prompt diagnosis care, patients frequently experience impairment or even pass away.  The  diagnosis is directly dased on the experience of the many doctors, and the situation is made worse by the numerous issues related to heart disease that place a great burden on them.  Therefore, it makes sense to introduce computer-aided approaches to help doctors diagnose cardiac disease in order to improve treatment.  Nowadays, researchers typically use the feature selection approach to the processed (13 features) dataset that was chosen by physicians. This is improper because the feature size is so small.  The usefulness of the unprocessed dataset is overlooked, and many are unaware that it may contain latent. The mRMR is better than previous approaches, and the incremental feature selection method works well. It has the least helpful features in addition to the best accuracy. On the Cleveland dataset, it has 100% accuracy with 8 features, on the Hungarian dataset, it has 98.3% accuracy with 14 features, and on the Long-beach-VA dataset, it has 99% accuracy with 9 characteristics. Additionally, we discover that certain characteristics—which physicians consider insignificant—have a role in classification and ought to catch their attention.