Demand forecasting and stochastic inventory management in blood banks

Many people need blood transfusions every day due to illnesses, surgeries or injuries. For this reason, it is vital problem that blood banks keep the right amount of blood. While keeping a small amount of blood in the blood bank stocks creates important problems such as not being able to meet the need and lossing the patients life, stocking a large amount of blood results in the deterioration of the blood and effecting the stock of different hospitals that need blood. Therefore, the optimum amount of blood should be kept in stock. In addition, the total cost of the blood bank is desired to be minimum. In this study, it is aimed to create a stochastic stock model that will predict the demand amounts of blood components at an optimum level and minimize the total cost of the blood bank. The developed model consists of five steps. The first step of the model was determining the variables that are affective in amount of blood component to be requested which was called collecting the data. Firstly, the blood component and transfusion centers were determined. In the second step, the amount of blood component is estimated using artificial intelligence estimation methods, such as ridge regression, gradient boosted tree, deep learning, decision tree, support vector machine, random forest and artificial neural network methods. In the third step, the best predicted demand estimation method was determined by statistical performance criteria. Demand forecast results; (s, Q), (s, S), (R, S) and (R, s, S) were used as input in stochastic stock models in the fourth step. Finally, in the fifth step, the stochastic inventory model which has the lowest cost is determined using developed cost model. Applicability of the developed model; it was examined in the Regional Blood Center, which provides blood components to public and private hospitals operating in Eskişehir, Afyonkarahisar, Kütahya and Bilecik. At the end of the application, it was seen that the model which has optimum excess and obsolete inventory stock quantities and minimum total cost is the (s, S) stochastic inventory model based on the support vector machine.