Approximate dynamic programming methods for advance patient scheduling

• Main Author: Sauré, Antoine
• Language: English
• Published: University of British Columbia 2012
• Online Access: http://hdl.handle.net/2429/43448

This dissertation studies an advance multi-priority patient scheduling problem. Patrick et al. (2008) formulated a version of this problem as a discounted infinite-horizon Markov decision process (MDP) and studied it using a linear programming method based on an affine value function approximation. This thesis starts by presenting an alternative solution approach for this problem based on the use of simulation, a policy iteration framework and a non-linear value function approximation. It then extends the dynamic multi-priority patient scheduling model and solution approach developed by Patrick et al. by considering patients who receive service across multiple days and for irregular lengths of time, and by allowing the possibility of using overtime on different days of the booking horizon. The research described in this dissertation is based on the hypothesis that some patients can be booked further into the future allowing the appointments for urgent patients to be scheduled earlier, and it seeks to identify effective policies for allocating available service capacity to incoming demand while reducing patient wait times in a cost-effective manner. Through the use of approximate dynamic programming techniques, it illustrates the importance of adequately assessing the future impact of today's decisions in order to more intelligently allocate capacity. Chapter 1 provides an overview of the multi-priority patient scheduling problem and a review of the literature relevant to it. Chapter 2 describes a simulation-based algorithm for solving a version of this problem and compares the performance of the resulting appointment scheduling policies against the performance of four other policies, including the one derived from the linear programming method. Chapter 3 extends the dynamic multi-priority patient scheduling model and solution approach developed by Patrick et al. It presents a discounted infinite-horizon MDP model for scheduling cancer treatments in radiation therapy units and a linear programming method for solving it. The benefits from the proposed method are evaluated by simulating its performance for a practical example based on data provided by the British Columbia Cancer Agency. Chapter 4 describes a teaching tool developed to illustrate advance patient scheduling practices to health care professionals and students. Finally, this dissertation concludes with additional discussion, extensions and further applications.