ICU Patients Classification System by Using CART Algorithm

Abstract

Nowadays, the advancement in computing technology and the realibility of computers has led to signification changes in the way that data are collected and analyzed. This system is to predict the risk level of patients in ICU (Intensive Care Unit) in time. Based on patients’ history records and current conditions, it classifies the patient’s risk level. As early warning scoring systems can improve outcome by detecting critical illeness earlier, the doctors in ICU can take care of patients effectively by using this system. Classification and Regression Tree (CART) use historical data with known classes to construct decision tree. CART use gini index impurity function. Then, it constructs the model or classifier to generate the from of classification rules. When new patients enter to the system, it classifies risk by using these rules. This system is implemented by using these rules. This system is implemented by using Microsoft Visual Studio 2005 and Microdoft SQL Server 2005.