Can Machine Learning Predict Mortality in Myocardial Infarction Patients within Several Hours of Hospitalization? A Comparative Analysis

Abstract

Cardiovascular Diseases, namely myocardial infarction (MI), is one of the leading cause of mortality globally. Despite all the medical advancements, more than half of MI patients have severe complications that go unnoticed or even untreated. In this study, we propose a Machine Learning (ML) powered framework to predict the deadly fate of MI patients. We trained various ML models to predict the lethal outcome following a myocardial infarction using a dataset of 1700 subjects and Ill clinical characteristics. Cox Regression was implemented to study the effect of various clinical phenotypes on the probability of patient survival. After preprocessing, sequential forward floating selector and recursive feature elimination were applied to select the right subset of the features for the various ML models. Numerous classification models were evaluated and optimized. The logistic regression classifier achieved an accuracy of 86.47% and a weighted F1 score of 86.92%.