Mixed Integer Optimization for Prioritized Surgery Scheduling

Operating room (OR) scheduling is one of the major budget cost sources and revenue generation part in health care system. Since general sequencing optimization problem on multiple ORs has been studied in numerous literature previously, this research proposes a novel strategy to solve the scheduling problem by introducing surgery urgency level in addition to operation cost. The objective of this paper is to get an optimal schedule by balancing four different OR operation costs with surgery priority predicted by multinomial logistic regression model.;In this research, surgery urgency is measured by priority level, which is highly related to surgery acuity level and delay days. To investigate the priority factors, multinomial logistic regression is utilized to predict surgery priority level using real-life based data set with information of surgery type, patient age, acuity level and the delayed days. Based on the priority prediction information, mixed integer programming (MIP) is used for building mathematical models of OR schedule. OR operation cost consists of four parts: basic surgery performance cost, overtime work cost, surgery cancel cost, and OR set up cost, which are set as multiple objectives for model construction. A priority-based OR schedule model is proposed with the consideration of OR operation costs and surgery urgency constraints in this research, which ensures high priority surgeries are scheduled prior to others in each OR under economical hospitals' budget. Three additional models are developed in this paper considering different objectives and constraints to compare with proposed priority-based OR schedule model. In model construction, Model I serves as the basic model; Model II includes OR set up cost in objective function on the basis of Model I; Model III adds surgery priority constraint based on Model I; and Model IV is the comprehensive proposed model which considers both surgery priority constraint and OR set up cost. Exact mathematical solution method is used to solve all developed schedule models in this research. Sensitivity analysis for multiple objectives is conducted by setting different parameters for all four models using weighted sum approach. Finally, optimal daily-based and weekly-based schedule lists are generated using the proposed priority-based surgery schedule model. Based on the weekly experiment results, higher priority surgeries are scheduled prior to others in each OR. Canceled surgery priority is increased in the following day's waiting list after re-evaluated from priority prediction model. The proposed integrated surgery schedule framework can be used on OR scheduling decision making for hospitals with massive surgery demands.