Predictive model for transferring stroke in-patients to Intensive Care Unit

Intensive Care Unit (ICU) admission is a main factor that affects the healthcare budget. Thus, the need for a predictive model for the decision to transfer stroke in-patients to the ICU is very important in order to utilize ICU resources effectively. Also, this predictive model will help to lower morbidity and mortality rates through earlier detection and intervention. This model could be used by an efficient clinical decision support system to assist healthcare professionals for faster decision-making. Currently, there is no research to predict the ICU transfer decision from vital signs of stroke in-patients. In this research, a Decision Tree (DT) model, an Artificial Neural Network (ANN) model, a Support Vector Machine (SVM) model, and a Logistic Regression (LR) model are evaluated for predicting the need to transfer the stroke in-patients to the ICU or not. The study is conducted on a clinical dataset consisting of 1,415 observations with six variables. The variables include temperature, respiratory rate, heart rate, systolic blood pressure (BP), oxygen saturations, and whether or not to transfer stroke in-patients to the ICU. The accuracy of DT, SVM, and LR are similar and equal to 0.96, whereas the accuracy of ANN is 0.94. Therefore, no specific model is better than others for making the decision. This is dependent on the nature of the dataset that was used for training and testing the models.