Informs Annual Meeting 2017

WC07

INFORMS Houston – 2017

WC07

WC08

322A Scheduling and Capacity Planning in Health Care Sponsored: Health Applications Sponsored Session Chair: Jonathan Patrick, University of Ottawa, 55 Laurier Avenue East, Ottawa, ON, K1N 6N5, Canada, patrick@telfer.uottawa.ca CChair: Antoine Sauré, University of Ottawa, 55 Laurier Avenue East, Ottawa, ON, K1N 6N5, Canada, antoine.saure@telfer.uottawa.ca 1 - Flexible Simulation Platform for Scheduling and Capacity Planning in a Radiotherapy Center Nadia Lahrichi, Ecole Polytechnique de Montreal, Cp6079 Succ. Centre-Ville, Montreal, QC, H3C3A7, Canada, nadia.lahrichi@polymtl.ca, Yosra El Abed, Louis-Martin Rousseau The objective of this work is to develop a flexible simulation platform that model several trajectories of patients in a radiotherapy center. Their interactions with resources are detailed and all processes are described. We use two practical cases to evaluate our platform and the literature to evaluate several organisation strategies: scheduling of patients, scheduling of resources and prioritization of tasks. 2 - Setting Patient Wait Time Targets in a Multi-Class Clinical Setting Vusal Babashov, University of Ottawa, 55 Laurier Ave E, Ottawa, ON, K1N 6N5, Canada, vusal.babashov@uottawa.ca, Antoine Saure, Jonathan Patrick In Canada, wait time targets for health care services are government-mandated. We believe that in the current setting patients wait longer than needed for no practical benefit to clinics in terms of the efficient use of resources. We aim to develop a methodology for determining the smallest possible wait time target for each patient class as well as to determine optimal regular-hour capacity for a clinic. To the best of our knowledge, this is the first study that attempts to determine appropriate wait time targets and capacity requirements simultaneously in a multi-priority clinical setting. 3 - Capacity Planning for a Network of Community Care Services Jonathan Patrick, University of Ottawa, 55 Laurier E, Ottawa, ON, K1N 6N5, Canada, patrick@telfer.uottawa.ca Community services have become increasingly important to health care delivery as people live longer but with chronic disease. We developed a queuing network model with blocking to determine the blocking probabilities within a network of services. We then utilize these to derive an optimal capacity plan through a simulated annealing approach. Results are verified through simulation. Transient behaviour of the system as capacity is adjusted to the optimized allocation is also assessed through simulation. 4 - Improving Access to Radiation Therapy Treatment Through Enhanced Patient Appointment Scheduling Claire Ma, MSc, British Columbia Cancer Agency, Vancouver, BC, We develop a decision support framework for radiation therapy appointment scheduling consisting of two parts. The first part assigns treatment dates and units to incoming patients, using policies derived from a discounted infinite horizon Markov model. The second part then assigns specific appointment times within each treatment date, using a mixed integer programming model. This framework provides a systematic way of allocating treatment capacity to incoming demand while improving patient satisfaction and service levels in a cost effective manner. The benefits of the proposed approach are evaluated by simulating its performance in a practical scenario based on operational data. 5 - Dynamic Advance Patient Scheduling with Resource Compatibility Restrictions Antoine Sauré, Assistant Professor, Telfer School of Management, University of Ottawa, 55 Laurier Avenue East, Ottawa, ON, K1N.6N5, Canada, antoine.saure@telfer.uottawa.ca, Ingeborg Bikker, Xiang Claire Ma We consider an advance patient scheduling problem in which patients require one or multiple appointments with one of several medical resources. We formulate the problem as a discounted infinite-horizon Markov decision process and solve it using approximate dynamic programming techniques. The main purpose of this model is to identify good policies for allocating available service capacity to incoming demand, while reducing patient wait times in a cost- effective manner. The benefits from the proposed method are evaluated by simulating its performance for two practical examples based on data provided by the British Columbia Cancer Agency. Canada, Claire.Ma@bccancer.bc.ca, Ingeborg Bikker, Nathaniel Horvath, Antoine Sauré, Scott Tyldesley, Aneeta Kassam, Martin L. Puterman

322B Learning Contributed Session Chair: Waldyn Martinez, Miami University, Oxford, OH, United States, martinwg@miamioh.edu 1 - Learning Travel Times in Traffic Networks under Noisy Observations Emily A. Meigs, MIT, 70 Pacific Street, Rm 243, Cambridge, MA, 02139, United States, emeigs@mit.edu We study the effects of unknown congestion based latency functions in road traffic networks on how users act over time. Specifically, we look at the convergence of knowledge under noisy observations in a nonatomic routing game. We characterize under what reasonable conditions and learning models will the users in the system converge to the full information Wardrop equilibrium. 2 - An Analysis of Base Learning Algorithms for Ensembles Waldyn Martinez, Assistant Professor of Business Analytics, Miami University, 117 Country Club Dr., Oxford, OH, 45056, United States, martinwg@miamioh.edu Ensembles are hybrid models composed of classifiers generated by a base- learning algorithm. Current available research is very limited in comparing the advantages and disadvantages of using different base-learners. Practitioners in several areas including finance, medical research, marketing, among others use ensembles as predictive models, but are limited to using decision trees as the main base-learning algorithm, given that most commercial packages and open-source software rely on the use of decision trees as ideal base-learners. The proposed research is an extensive comparison of base-learners within the most commonly used ensemble methods. 330A Flexibility in Manufacturing and Service Operations Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Hoda Bidkhori, University of Pittsburgh, Pittburgh, PA, 15206, United States, bidkhori@pitt.edu 1 - Analysis of Process Flexibility Designs under Disruptions Erfan Mehmanchi, University of Pittsburgh, 1025 Benedum Hall, WC09 We examine the worst-case performance of flexibility designs under demand and supply uncertainties. Supply uncertainty is considered in the form of plant and link disruptions. In particular, we establish results that allow the decision-maker to compare the performance of different designs. We also discuss fragility of various designs that is defined as the impact of disruptions on their performance. 2 - Stability in Multi Professional and Multi Skilled Services Hoda Bidkhori, University of Pittsburgh, Dept of Industrial Consider a system where we have different types of multi professional skilled workers. Each professional worker can be flexible to do different tasks which require different skills. Training multi skilled professional workers is costly. We first discuss the optimal flexibility and the learning in such system 3 - Fair Implementation of Interruptible Demand Response Programs Ali Fattahi, UCLA, Los Angeles, CA, United States, ali.fattahi.1@anderson.ucla.edu, Sriram Dasu, Reza Ahmadi Direct load control programs are very popular among utility companies for curtailing electricity consumption during peak load periods. These programs require that the participating customers reduce their consumption, by a predetermined amount, a few times during a year. We study fairness in the assignment of interruptions to customers. Engineering, Pittburgh, PA, 15206, United States, bidkhori@pitt.edu, Sriram Dasu, Reza Ahmadi 3700 O’Hara Street, Pittsburgh, PA, 15261, United States, ERM83@pitt.edu, Hoda Bidkhori, Oleg A. Prokopyev

506