PREDICTION OF DEMAND FOR RED BLOOD CELLS USING RIDGE REGRESSION, ARTIFICIAL NEURAL NETWORK, AND INTEGRATED TAGUCHI-ARTIFICIAL NEURAL NETWORK APPROACH

Abstract

Regional blood centers collect donated blood and distribute processed blood to the blood transfusion centers according to their need. The prediction of blood components to be demanded by transfusion centers becomes more important, especially these days when the impact of COVID-19 is increasing. Since donors are afraid to go to blood donation centers, blood component stocks rapidly decrease. This study aims to predict the blood transfusion centers' demand for quantities of red blood cells, which is an important blood component, from a regional blood center by using the artificial neural network method. The method's parameters values affect the prediction performance of the method. Therefore, the Taguchi method is integrated with artificial neural network methods to optimize the parameters. The prediction results of the integrated Taguchi-artificial neural network approach, artificial neural network method, and ridge regression method are each compared with the actual demand of regional blood centers. It is determined that the integrated Taguchi-artificial neural network approach predicts actual demand more accurately.