具回流性質之醫療系統動態資源配置

Abstract

本研究即探討具有回流性質的等候系統之允入機制，醫療產業中之住院部門即為一具回流性質系統，病患出院後可能會於短時間內再次住院，因此對於醫療機構而言有兩類不同的病患需求，初住院病患與再住院病患。再住院病患的病情通常較初次住院的病患來的嚴重，需要立即的照護，因此基於人道立場醫院應保留資源給潛在再住院病患，然而保留太多或太少皆會對整個醫療系統產生負面影響。 本研究即以動態規劃為方法，提出一動態的病患允入機制，考量兩類需求來到、服務率等不確定性，於各期不同的狀態下求得最佳允入水準。並提出回流醫療系統病患允入模型(Re-entrant Healthcare System Inpatients Admission, R-HSIA)，其為一啟發式動態允入模型，用以趨近動態規劃之最佳解。為驗證RHSIA模型求解後所得策略之可行性，我們建構一模擬醫院允入控制流程程式，並與其他常見之醫院病患允入策略(先到先服務法FCFS、各科各自決策法SE)進行實驗設計，將本研究求得之R-HSIA允入決策與模擬結果比較分析。模擬結果說明本啟發式模型，在醫療供給與需求相近時啟發式允入模型效用優於其他兩種方法，並可見其之可行性與穩健性。另外，本研究所提出的R-HSIA模型不但能處理較複雜的資源分配問題，同時也能加快求解速度。最後我們也進行程式模擬來驗證模型的正確性與可行性。

This research studies flexible resources allocation problem in re-entrant healthcare systems. For healthcare service, inpatient department is an example of re-entrant system. Inpatients may return shortly after original discharge from the hospital. Therefore, for the inpatient department, there are two types of demands: initial admission and readmission patients. In practice, readmitted patients are usually sicker than initial patients. Thus, for humanity consideration, hospital should give readmission patients higher priority. However, the real demand of readmission patients is usually unpredictable. In order to reserve resources for potential readmission demand, hospital might need to control the number of initial admission patients. Nevertheless, either too much or too little resource reservation has negative impact on healthcare institute. Too much reservation may generate higher resource idle cost, and too little reservation may cause higher rejection cost of readmission patients. To solve this resource allocation problem, it is necessary to develop an efficient admission control mechanism for healthcare service providers. In this research, we propose a dynamic admission control model to allocate a central pool of resources between initial admissions and readmissions patients. We also consider uncertainties of initial admission demands, readmission demands, and number of high risk discharges in the model. At the beginning of each period, we have to decide the number of initial patients can be admitted. By adopting a stochastic dynamic programming approach, we obtain an optimal admission policy that can achieve minimal total cost in the long term. At last, we develop a heuristic model, re-entrant healthcare system inpatients admission model, as a simpler alternative to the optimal policy. Our models are validated in extensive simulation study.