Informatics in Medicine Unlocked (Jan 2019)
Prediction of mortality associated with early onset pneumonia in Acute Myocardial Infarction
Abstract
Acute Myocardial Infarction (AMI) is a serious heart condition, affecting about 1 million Americans each year. AMI is generally followed by an emergency revascularization to restore blood flow and minimize heart damage; however, even when the procedure is successful, patients are still prone to infectious diseases, such as pneumonia. The objective of this study is to develop a mathematical model to predict in-hospital mortality caused by Early Onset Pneumonia (EOP) in patients with AMI. We identified and collected 14 clinically important parameters from 101 EOP patients. Using logistic regression analysis, we developed a prediction model with three statistically significant parameters: diabetes, low left ventricular ejection fraction (LVEF) (<40%), and contrast induced nephropathy (CIN). Our model has an AUC of 0.9, sensitivity of 70.2%, specificity of 92.2%, and an overall accuracy of 87.1%. Based on these three parameters, we further presented a checkout table, which can be easily used by a nurse or a physician for quick evaluation of the level of mortality risk for AMI patients with EOP. The results of this work may provide insights to improve the quality of care and reduce the cost for EOP treatment in AMI. Keywords: Acute myocardial infarction, Early onset pneumonia, Mortality prediction, Multivariate regression, Predictive model