The Inventory Policy on Regional Blood Center

Abstract

Blood is not only an indispensable medical resource in the hospital but also a perish product with a limited shelf life which will reduce the quality as the storage time shortens It may cause unnecessary waste of social resources or increase the risk of patient treatment Across all blood types the shelf life of platelets is only five days after donation which shows the importance of inventory management In the past most research used the Taguchi method to solve the optimal parameters and considered only one inventory policy We propose a model combining the neural networks and simulation optimization not only taking the trend of platelet demand into consideration but also discussing multiple inventory policies Firstly we use Long Short-Term Memory (LSTM) to predict the demand of the hospitals and use simulation optimization to find the optimal parameters Secondly considering the predicting demand of hospitals and the optimal parameters simultaneously we use LSTM to predict the corresponding parameters This model was tested in six scenarios and we used the number of shortages expiration total inventory cost and average remaining life to evaluate the performance Finally a numerical example is presented to demonstrate this model accompanied by sensitivity analysis The results in all scenarios performed better than the baseline and the best performance was when there is was high donation limit The number of shortages is lower when using the order-up to policy and periodic inventory system The amount of expired blood is lower when using the (s Q) policy Some managerial insights are obtained from the results which will assist the blood center in improving its inventory management

Date of Award	2020
Original language	English
Supervisor	Tai-Yue Wang (Supervisor)