| 摘要: | 良好的醫療物資輸配送作業規劃，可幫助醫療系統於緊急疫情爆發時，有效率地服務突然大增的病患物資需求，以確保醫療系統物資資源有效率地應用與維持一定的醫療服務品質。目前實務上醫療院所之物資訂購作業上，係由醫療系統成立之聯合決策中心搭配決策支援系統進行規劃，但對於醫療物資訂購/配送頻率、訂購量及安全存量等重要參數值的決策，仍依靠人工的經驗來制定。此作法不具系統性分析，且相當依賴相關作業人員的主觀判斷，因此常可能導致不佳的決策。尤其，在疫情爆發時，物資需求可能突然大量增加，可能導致原先規劃的物資配送結果難以運用於實際作業中，且現有的決策作法不易有效地調整現有的資源以因應擾動事件。另外，在實際的營運中，醫療物資需求常會隨著的許多隨機事件產生擾動變化，而可能導致原先規劃的結果失去其最佳性，進而降低整體醫療系統物資的應用成效，造成醫療系統成本的大幅增加。有鑑於此，本研究考量疫情發生時預期各時點的物資需求、各醫療院所各時點的存貨容量限制、物資運送量及其他相關限制，並探討物資在時間面與空間面之整合及相互支援作業，建構一確定性醫療物資輸配送模式。再進一步針對於營運時，醫療物資需求量產生的隨機擾動現象，構建一隨機性醫療物資輸配送模式。本研究模式期能在未來實務的應用上，提供醫療系統在緊急疫情爆發時，有效地規劃醫療物資訂購及輸配送作業。 本研究利用時空網路流動技巧構建確定性模式，此模式含有多個時空網路，以定式不同類醫療物資在時空中的流動情況，可定式為一整數多重貨物網路流動問題，屬NP-hard問題。本研究進一步修正確定性模式中之固定需求值為隨機需求值，建立隨機性模式。注意，由於此隨機性模式之問題規模與隨機事件數相關，因此其規模將遠大於確定性模式規模，為更有效求解此隨機性問題，本研究進而發展一以問題分解策略為基礎之啟發解法，以有效地求解隨機性模式。此外，為比較確定性與隨機性模式所求得之結果，在隨機營運環境中的績效優劣，本研究發展一評估方法。最後，為測試模式與求解演算法於實務營運之績效，本研究以國內一大型醫療系統的營運資料為例，進行實例測試與分析，結果甚佳，顯示本研究所建構之模式與求解方法應可為未來實務醫療管理當局之參考。 When an urgent epidemic situation breaks out, a satisfactory schedule of medical resource supply orders and transit plan can help a medical system efficiently serve drastically increased demand of medical goods to patients. It can also help effectively reduce the operating cost and maintenance the medical service quality. Currently medical goods are ordered by the union decision center in a Taiwan medical system with electronic purchase systems and decision support systems. However, the important parameters (e.g., the order/transit frequency, the order quantity, and the safe stock capacity) are manually determined by staff with experience. Lacking a systematic optimization analysis, this approach rather depends on the staff’s subjective judgments. As a result, feasible but inferior decisions have usually been made. In particular, under an urgent epidemic situation, the demand of medical goods would suddenly and largely increase, which would make it difficult to efficiently revise the original schedule with existent resources to respond to the incident. Additionally, in actual operations, the demands of medical goods often change stochastically, possibly causing the original schedule to lose its optimality. Consequently, the effect of medical system would be decreased and the operating cost be increased. Therefore, in this research, based on a medical system’s perspective, we systematically consider the expected epidemic period, the expected demand of goods for every time slot in all hospitals and their departments, the transportation cost of goods, the stock capacity and other constraints, as well as the integrated transit plan of medical goods in the dimensions of time and space, to construct a deterministic medical goods order and transit scheduling model. Further, considering the stochastic demand of medical goods that occur in real time operations, we construct a stochastic medical goods order and transit scheduling model. These two models are expected to be useful planning tools for medical system to determine effective resource supply orders and transit schedules under urgent epidemic situations. We employed time-space network techniques with the system optimization perspective to construct a deterministic real-time scheduling model, which include several time-space networks to express the flows of different medical goods in the dimensions of time and space. The model is formulated as a multiple commodity network flow problem that is characterized as NP-hard. Then, we modified the fixed demand parameters in the deterministic scheduling model as random variables to develop a stochastic real-time scheduling model, which is more complicated than the deterministic scheduling model. To better solve the stochastic model, we used problem decomposition techniques to develop the heuristic to solve the stochastic problem. In addition, to evaluate the deterministic and stochastic scheduling models in actual operations, we develop an evaluation method. Finally, in order to test the models and solution algorithms in actual operations, we perform a case study based on a domestic large-scale medical system (containing several hospitals)’s operating data. The preliminary results are good, showing that the models could be useful for medical system planning medical goods order and transit scheduling. |
