Mariia Anapolska: Robust Appointment Scheduling in Hospitals
As the demand for health care services increases each year, the need for efficient management of health care systems becomes more and more apparent. One of the most important health care providers are hospitals. Hospitals are under tremendous cost pressure and must achieve a balance between economic efficiency and a treatment that focuses on the patient. To improve clinical operations and patient safety, my research considers the appointment scheduling problem within a hospital. Problem description. The problem aims to maximize the utilization of the hospital resources while minimizing the patients’ inconveniences such as waiting time. Typically, an arriving patient needs to undergo several types of treatment. This means that several hospital resources will be needed either simultaneously or sequentially in a short time period. The treatments must be scheduled so that they satisfy the resource capacity restrictions. The hospital environment is very dynamic: The length of patients’ treatments varies and arriving patients represent an uncertain demand for resources. The presence of emergency patients requires the schedule to be highly adaptable, i. e., robust and stable solutions are needed. Envisioned work. Solutions of robust optimization problems depend on the uncertainty sets constituting the problem’s input. In robust optimization, researchers assume these sets to be given by experts. However, experts often do not understand the dynamics within robust optimization, e.g., that integrating scenarios with high fluctuations leads to unpredictably high costs. Furthermore, especially in the hospital context, even for experts it is quite difficult to measure and obtain all data needed for presenting a scenario. To overcome this obstacle, we will use agent-based simulation to obtain all important parameters. To that end, the simulation framework “SiM-Care” developed by Martin Comis needs to be extended and adapted. This agent-based simulation models interactions between the population and the physicians in a primary care system. It evaluates the input health care system by computing performance indicators that characterize the system’s efficiency both from patients’ and physicians’ points of view. Moreover, the simulation allows us to assess the impact of changes in the system, such as changes in the patient-to-physician ratio or novel management strategies of physicians. In order to obtain realistic input scenarios for the appointment scheduling problem, we plan to extend the model of SiM-Care further in order to integrate emergency and elective patients requiring hospital treatment. Since SiM-Care produces scenarios based on parameterized probability distributions, we will investigate the influence of the uncertainty sets for demands generated by SiM-Care on the resulting solutions for the robust appointment scheduling problem.