區域血液中心之存貨政策分析

Analysis on the Inventory Policies of Regional Blood Centers

研究目的：本研究針對區域血液中心的血液存貨進行血品壽命與成本分析，同時引入人工智慧方法進行血液需求預測與分析，並比較不同血液存貨管理方法的優劣。研究方法：基於血液的需求及供給皆是不確定性的，會隨著時間或事件的發生而波動，形成一時間序列，本研究以血小板為主要研究對象，利用人工智慧中長短期記憶網路預測未來血小板的供應量，接著以模擬最佳化的方法求解該期最佳存貨策略參數，最後再利用類神經網路模型，預測該期之最佳存貨策略參數，提供血庫管理人員控制血液捐贈量。研究結果：研究發現，在區域血液中心的存貨管理上，血液缺貨及血液過期方面，本研究所提出的法都優於現有的管理方法；並發現當缺貨數量較多時，供應的血小板平均剩餘壽命較高，當過期數量較多時，平均剩餘壽命較低。研究限制：本研究使用區域血液中心的資料作為實例驗證，在不同區域血液中心或擴大為全國性血液調度時，可能會產生效益上的差異。實務意涵：本研究提出之方法提供血庫管理人員能夠明確控制血小板的存貨，依據管理需求採取不同的存貨策略，並將過期處理、缺貨及存貨持有成本降到最低，甚至能夠將此方法應用至其他血品之存貨管理。價值：從區域血液中心的角度來理解血液存貨管理的意涵，尤其是對於短天期的易腐性存貨管理，正確的預測更能顯示出存貨管理的價值。

Purpose - This study aims to, with regard to the blood inventories of regional blood centers, conduct blood shelf life and cost analyses and import artificial intelligence methods to conduct blood demand forecasts and analysis, as well as compare the pros and cons of different blood inventory management methods.Design/ methodology/ approach - Due to the uncertain nature of the demand and provision of blood, which, in a sense, can fluctuate depending on the time or occurrence of certain events, a time series has been formed. This study used platelets as the primary research subject and adopted the long short-term memory (LSTM) network of artificial intelligence to forecast the future supply of platelets. Subsequently, simulation optimization methods were used to determine the optimal inventory policy parameters for the given period, after which the neutral network model was once again used to forecast the optimal inventory policy parameters for the given period. This was done to provide to blood inventory management personnel with a means to enable them to control the amount of donated blood.Findings - The research discovered that, with regard to the inventory management of regional blood centers, the management methods proposed by this study in the aspects of blood inventory shortage and blood expiration are superior to the methods currently adopted. In addition, during periods of relatively severe blood inventory shortage, the average remaining shelf life of platelets provided was relatively high, while the average remaining shelf life was low when the expired quantity was relatively large.Research limitations - This research used the data of a single regional blood center as an empirical example, which, if applied to different regional blood centers or even extended to national-level blood dispatch, could produce differences in effectiveness.Practical implications - The methods proposed by this research provide blood inventory management personnel with clear methods for controlling the inventory of blood platelets, helping them to adopt different inventory policies according to management needs as well as to minimize expiration processing, shortage, and inventory holding costs. Moreover, this method could even be used for the inventory management of other blood products.Value - In understanding the implications of blood inventory management from the perspective of regional blood centers, especially with regard to short-term perishable inventory management, the more accurate the forecast, the greater the inventory management values proved to be.