Robust Optimization in Healthcare Management
醫療管理中鲁棒優化的應用
Student thesis: Doctoral Thesis
Author(s)
Related Research Unit(s)
Detail(s)
| Awarding Institution | |
|---|---|
| Supervisors/Advisors |
|
| Award date | 31 Aug 2017 |
Link(s)
| Permanent Link | https://scholars.cityu.edu.hk/en/theses/theses(bc83fe51-2910-486c-8dc5-c3f56ef05727).html |
|---|---|
| Other link(s) | Links |
Abstract
Governments in most countries are struggling to enhance the efficiency of their healthcare services and systems while reducing the costs of healthcare. The challenges are presented due to the booming population growth, urbanization of cities, growing life expectancy and so on; however, the limited resources such as money and land restrained its capability to accommodate such substantial trends. In order to overcome various possible challenges in the future, hospitals have to improve the efficiency of the existing operations amidst high uncertainty on information such as demand and service time.
In this thesis, robust optimization method is recommended for embracing the uncertain data such as demand fluctuation and service time uncertainty. This method reduces the computational time for hospital decision makers and many existing software packages are available for calculation. Furthermore, the distribution is not required to be assumed as it is in the traditional methods. It can be applied to three main problems in hospital: 1) appointment scheduling with overbooking strategy 2) capacity planning in Ambulatory Care Center and 3) bed management for inpatient admission.
In the first problem, hospital managers have to determine the number of patients to accommodate for consultation at different time. Robust linear programming model is adopted for solving the problem with the fluctuating number of no-shows and demand. Excessive number of invited patients will cost higher overbook compensation while too few will cause idle utilization of the resources, which make the whole system inefficient. As for the second problem, the rooms to be opened have to be decided for a certain period of time. Under the stochastic service time and demand, a robust mixed integer programming is employed to identify the rooms to be opened and worth investing in. In the third problem, the number of beds reserved for regular inpatients have to be tackled by hospital managers. Two robust optimization models are presented for bounded uncertainty and symmetric uncertainty of bed demands. For these three problems, numerical examples are demonstrated to indicate the effectiveness of the models. Finally, a conclusion of the thesis is drawn and further research directions are outlined.
In this thesis, robust optimization method is recommended for embracing the uncertain data such as demand fluctuation and service time uncertainty. This method reduces the computational time for hospital decision makers and many existing software packages are available for calculation. Furthermore, the distribution is not required to be assumed as it is in the traditional methods. It can be applied to three main problems in hospital: 1) appointment scheduling with overbooking strategy 2) capacity planning in Ambulatory Care Center and 3) bed management for inpatient admission.
In the first problem, hospital managers have to determine the number of patients to accommodate for consultation at different time. Robust linear programming model is adopted for solving the problem with the fluctuating number of no-shows and demand. Excessive number of invited patients will cost higher overbook compensation while too few will cause idle utilization of the resources, which make the whole system inefficient. As for the second problem, the rooms to be opened have to be decided for a certain period of time. Under the stochastic service time and demand, a robust mixed integer programming is employed to identify the rooms to be opened and worth investing in. In the third problem, the number of beds reserved for regular inpatients have to be tackled by hospital managers. Two robust optimization models are presented for bounded uncertainty and symmetric uncertainty of bed demands. For these three problems, numerical examples are demonstrated to indicate the effectiveness of the models. Finally, a conclusion of the thesis is drawn and further research directions are outlined.
