Reporting time: Tuesday ,13 October 14:00-15:30 p.m
Tencent Conference Number :155 316 805 Password :123456
Speaker: Professor Shu Jia, Vice-President, School of Economics and Management, Southeast University
Speaker Profile:
Shu Jia, Professor, School of Economics and Management, Southeast University, research field: logistics, supply chain management; traffic management; medical management.
Introduction:
Mobile operating theaters (MOTs) emerged as a solution in coping with the serious shortage of medical resources in various European countries. In this paper, we study the multi-MOT elective surgery planning problem. Each MOT consists of an operating room (OR) and a recovery room, i.e., post-anesthesia care unit (PACU), with two beds, which can be used to perform surgeries and recovery cares for both elective and emergency patients. The problem is to determine the daily assignment of elective surgeries to MOTs under uncertain demand of emergency patients in a finite planning horizon. The objective is to minimize the total daily costs of the elective patients and MOT assignment and the MOT utilization as well as postponement cost caused by the delay of any elective surgery to the next planning horizon. Due to the possible OR blocking in the presence of limited PACU capacity, we cannot obtain the daily MOT utilization cost in an analytical form in contrast to the existing literature. The daily MOT utilization cost is separated into the components of daily OR and PACU utilizations that are assumed to be general functions of the total surgery and recovery durations of the elective patients, respectively. We structure the problem as a set-partitioning problem and solved it by column generation. We show that the pricing problem that arises from the column generation algorithm is NP-hard. By effectively characterizing the structural properties of the optimal solution to the continuous relaxation of the pricing problem, an efficient implementation of the branch-and-bound procedure can be applied to obtain the optimal integral solution to the pricing problem. For the purpose of implementation, we learn the cost functions of daily OR and PACU utilizations with the data randomly generated according to the associated empirical distributions documented in the literature. Computational results demonstrate that the proposed approach can solve moderate-sized random instances very effectively. Computational results also show that comparing with the traditional OR planning model that does not consider the daily cost of PACU utilization and thereby ignores the possible idle and blocking times between surgeries in ORs, our model results in a total cost reduction of 20%. This is a joint work with Zhang Minghui and Zhu Zhicheng.
