探討季節性流感防疫於不確定性環境下疫苗供應鏈策略之分析

Abstract

季節性流感為台灣一重要的議題，每年超過10%的人口感染，除了影響國民健康之外，也造成國家經濟上的損失，政府每年推動疫苗接種計劃，訂購約300萬劑公費流感疫苗，針對高風險感染族群，提供免費接種服務，以防止病情擴散。然而，在防疫計劃的執行面上，仍有許多供應鏈的問題待被解決。本研究論文著重在兩個研究主題上，首先，因政府建議共享疫苗類型或再分配辦法以減緩疫苗短缺之危急，但目前為止，鮮有相關文獻或計量方法探討該政策之優劣，因此，本論文將建構一隨機存貨模型，分析疫苗存貨共享策略之經濟效益。根據研究結果，我們發現存貨共享策略可有效降低成本，其中，又以共享疫苗類型的經濟效益較大，且幼兒型疫苗可有效降低成本的效果較成人型疫苗更為明顯。再者，推動流感疫苗接種計劃，並非以達群體免疫門檻為目標，而為降低因預防及感染流感所造成的整體花費，因此，我們發展一兩階段隨機規劃模型，探討疫苗訂購與分配決策間之交互影響，強化供應鏈協作，以最佳化醫療資源的利用。本模型第一階段之決策，將求解成人型與幼兒型流感疫苗訂購量，第二階段決策基於第一階段之求解結果，將疫苗效力、流感傳染率、及疫苗需求等不確定性因素納入考慮，擬定各型疫苗在不同群體間之分配策略，與可用於緊急需求之總量。本研究建構自動化程序，可批次求解真實世界之複雜情境。研究分析採納真實數據資料，顯示隨機規劃模型之求解結果，相較於確定性模型的表現更為穩定，同時，亦指出不確定性因素對於防疫成效具明顯影響，本論文之研究成果預期可協助國內相關醫療決策訂定，提升流感防疫成效。

Seasonal influenza outbreak has been a grave issue in Taiwan. Over 10% of populations are infected by influenza viruses, and the epidemics have caused loses on both health and economics. Recently, the government has implemented vaccination campaigns to prevent the diseases by providing 3 million doses of free vaccines to the high-risk populations in each year. However, there are still issues in the vaccine supply chain that post a threat for the implementation of the immunization program. This thesis focuses on two research topics. We first formulate a stochastic inventory model to investigate the cost effectiveness of vaccine inventory pooling policies. This study is motivated by the government recommendations of sharing and redistributing vaccine inventories when a shortage occurs. To the best of our knowledge, there is no such research focusing on this specific topic. The results show that the sharing policies reduce costs. In particular, the policy of sharing adult and pediatric vaccines has more cost reductions than the centralized policy. Our findings suggest that the pediatric vaccine obtains more savings than the adult vaccine through sharing and redistribution policies. In the second study, we develop a two-stage stochastic programming model to investigate the end-to-end decisions in a vaccine supply chain. Instead of considering to the herd immunity, the objective is to minimize the overall economic impact of prevention and treatment caused by the influenza. The first-stage decisions are to determine the ordering quantities of both adult and pediatric vaccines, and the second-stage decisions are to allocate vaccines among age-group populations and to determine the expedite requirements. Various uncertain factors are considered including vaccine efficacy, infection rate and demand. We implement an automatic process to obtain solutions from a large-scale setting of scenarios from the real-world data. The stochastic model provides us a more robust solution than the model only considering to a deterministic parameter verified by the computational results. By using the epidemic data, our findings offer valuable insights for decision makers across government, healthcare providers and other organizations to impact to the operational design of healthcare process.