Stepwise Regression Machine Learning Models for In-Hospital Mortality Prediction in Patients After ST-Segment Slevation Myocardial Infarction (STEMI)

Abstract

Acute myocardial infarction is a leading cause of cardiogenic shock and mortality. The aim of current study is to identify factors and develop machine learning models that predicts in-hospital mortality of ST-segment elevation myocardial infarction (STEMI) patients in South-East Asian population. This is a single center, retrospective study, from patients presenedt to the emergency room at Kaohsiung Chang Gung Memorial Hospital, Taiwan. The study included non-trauma adults (≥20 years of age) who were diagnosed with acute STEMI. A total of 1567 patients who met the inclusion criteria were enrolled. The area under the receiver operating characteristic curve was 0.839 in logistic regression (LR) and 0.825 in random forest (RF). The accuracy was 0.821 in LR and 0.812 in RF. The sensitivity and specificity were 0.883 and 0.815 in LR, and 0.875 and 0.806 in RF. In conclusion, the predictive models developed using LR and RF algorithms can be used to predict the risk of in-hospital death for STEMI patients.