Track: Healthcare Operations and Services

Abstract

Studies have shown that the mortality rate due to ST-segment elevation myocardial infarction (STEMI) has drastically increased in developed and developing countries. The purpose of this study is to develop a diagnostic support tool to help classify STEMI patients and validate the predictive risk factors associated with STEMI using an ensemble learning approach. In this retrospective data-mining study, the data are retrieved from electronic health records of an urban emergency department between January 2017 and August 2020. A Random Forest model is trained to classify non-acute coronary syndrome (non-ACS) etiologies and STEMI patients using 38 features. Of the study cohort, 411 patients with chest pain fulfilled inclusion criteria of whom, 225 (55%) are STEMI, and 186 (45%) are non-ACS etiologies patients. The proposed framework successfully classifies the non-ACS etiologies and STEMI patients with recall and area under the receiver operating characteristic (AUROC) values of 73% and 90%, respectively.