Minimum Skewness based Myocardial Infarction Detection Model using Classification Algorithms

Abstract

Myocardial infarction is one of the most dangerous public health issues in the world. The accurate prediction of myocardial infarction disease aids in disease diagnosis and biological analysis of the patient's health. The classification algorithms are one of the solutions that predict accurate diseases based on the symptoms (attributes) in patients' details. The ability to predict accurately reduces the risk of causality and decision-making time. This study proposed the Minimum Skewness-Based Myocardial Infarction Detection Model (MSMIDM) with the help of a statistical and feature selection-based approach. Minimum skewness is a feature selection statistical approach that selects essential attributes of a dataset. The MSMIDM provides accurate results with the highest accuracy among the six classification algorithms. The experimental analysis makes use of the most widely used Cleveland dataset for myocardial infarction detection. The experimental results are analyzed through a confusion metric, statistical, and partitional validation approach. The proposed model obtains an accuracy of 90%, 87.037%, 87.037%, 83.238%, 81.481%, and 85.556% with respect to Random Forest, K-Nearest Neighbor, Support Vector Machine, Naive Bayes, Decision Tree and Neural Network classification algorithms. According to the experimental analysis, this study suggests the MSMIDM-based Random Forest algorithm is excellent for myocardial infarction disease detection.