期刊名称：Journal of Minimally Invasive Surgical Sciences
印刷版ISSN：2251-7022
电子版ISSN：2251-7030
出版年度：2013
卷号：2
期号：2
页码：129-137
出版社：Kowsar
摘要：Background: Operating room (OR) planning involves the creation of a master surgical schedule in which surgeons are assigned to specific operating rooms (ORs) on specific days of a week. The master schedule is typically one or two weeks long repeatable for several months. Objectives: The purpose of this study was to recommend using a mathematical program to generate a rotation in a way that the limited operating room capacity could be distributed based on smoothing expected demand for in-patient beds. Patients and Methods: This study concentrated on the service-level scheduling at Sunnybrook Health Sciences Centre in Toronto, Canada, to build such a model. We assumed that the number of blocks (days) for each surgeon was given, and that the expected case-mix for each surgeon was chosen by random sampling based on historical data. The goal was to assign surgeons to the blocks so tat bed occupancy in the wards would become as stable as possible during the week. The planning problem was first formulated as a stochastic integer programming. Then, an approach with combination of Monte Carlo simulation and Premium Solver provided an approximate solution. Results: The integer program provided scheduled OR number and day of the week for each surgeon, corresponding to the sample. The final result of model, approximated by the proposed method, was the maximum number of beds for each surgical service throughout the week. These were the required bed capacities to handle demands for surgeries. Conclusions: An Integer Programming was presented to schedule OR and day of surgery for each surgeon with restrictions on the available ORs and required number of blocks. The problem was quickly solved using Premium Solver. The reliability of the results was highly dependent on the data. Another fundamental restriction for implementation of the results was to convince surgeons to accept changes in the schedules. The surgeon preferences might be included in the model constraints for more acceptable results.
