Mortality prediction of ICU patients using EDA-enhanced logistic model

Due to the different health conditions of an increasing number of serious patients, the Intensive Care Unit (ICU) of a hospital has to correctly classify patients according to their conditions so that medical resources could be properly utilised. The seriousness of the illness can be classified based on the significant risk factors and its corresponding impacts on the patients’ survival. How to quickly identify the significant variables is a major task for classification. This paper proposes a Multistage-EDA-Enhanced Logistic Regression (MEDAeLR) approach to precisely classify the patients and quickly diagnose with three-stage analysis. A cohort of 200 consecutive ICU patients was borrowed for validation. Regular MLR, classification trees and Linear Discriminant Analysis (LDA) are carried to compare the performance with proposed method. The results show that MEDAeLR provides more satisfactory identification performance in terms of Receiver Operating Characteristic (ROC) curve and Area under the ROC Curve (AUC).