Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MB60

Gregory Critchley, Western University-Ivey Business School, 10 Blosdale Circle, Delaware, ON, N0L1E0, Canada, Lauren Cipriano, Greg Zaric, Jeremy D. Goldhaber-Fiebert Multi-payer health care systems are common. For example, in the United States, many individuals are covered under private or employer-sponsored health plans prior to age 65 and are thereafter covered under Medicare. We model a multi- payer health care system using a Markov decision process and show that the penultimate payer, within a patient’s lifetime, will prescribe treatment sub- optimally. We find that there exists a coordinating transfer payment between payers that results in a socially optimal treatment plan. Using the hepatitis C virus in a numerical example, we quantify the impact of a non-coordinated multi-payer health care system. 2 - Sounding the Alarm on Opioids Margret V. Bjarnadottir, University of Maryland, 4353 Van Munching Hall, College Park, MD, 20742-0001, United States, David Anderson, Ritu Agarwal, Kenyon Crowley, Kislaya Prasad, Alan Nelson The ongoing opioid epidemic is a serious public health issue. In our paper, we investigate the feasibility of early detection of chronic opioid use and build advanced machine learning models that can be incorporated into clinical decision support systems, potentially minimizing adverse events associated with chronic opioid use and dependency. 3 - Appointment Access in Family Medicine Clinic Vera Tilson, University of Rochester, 3-343 Carol Simon Hall, W. E. Simon Graduate School of Business, Rochester, NY, 14627, United States, Ryan Spurr We discuss a scheduling approach to improve access in a family medicine clinic. 4 - Personalized Risk Management Strategies for Women at High Risk of Developing Breast Cancer and the Role of Adherence Caglar Caglayan, Georgia Institute of Technology, 755 Ferst Drive NW, Atlanta, GA, 30332, United States, Turgay Ayer, Kalyan Pasupathy, Sandhya Pruthi Women with BRCA 1/2 gene mutations or family history are at higher risk for breast cancer. The risk management interventions for high-risk women include intensified screening, preventive surgery (e.g., mastectomy) and risk-reducing medications (i.e., chemoprevention). Individual factors such as breast density and adherence behavior play a critical role at identifying and tailoring the optimal risk-management strategies for individuals at high-risk. In this work, we study breast cancer risk management problem with a comprehensive simulation model and identify optimal personalized strategies for high-risk individuals considering key patient characteristics and adherence behavior. n MB60 West Bldg 102B Managing Cost and Quality in Healthcare Sponsored: Health Applications Sponsored Session Chair: Ozlem Yildiz, University of Virginia, Darden School of Business, United States 1 - Fair Machine Learning with Health Data Mahbod Olfat, PhD Student, University of California-Berkeley, 1433 Dwight Way, Unit C, Berkeley, CA, 94702, United States, Anil Aswani The explosion of big data in health care has given warrant for concern about proliferation of bias and inefficient delivery of health care though machine learning methods. While researchers have begun to study this problem for supervised learning problems, it has been less well-explored under the paradigm of unsupervised learning. In this talk, we discuss fairness in the context of dimensionality reduction via principal component analysis (PCA). We present a definition of fairness and develop a convex SDP program that can achieve it. We conclude by showing how our approach can be used to perform a fair (with respect to age) clustering of health data that may be used to set health insurance rates. 2 - Does Mandatory Overtime Law Improve Quality? Staffing Decisions and Operational Flexibility of Nursing Homes Susan F. Lu, Purdue University, Krannert 441, West Lafayette, IN, 47907, United States, Lauren Lu During the 2000s, over a dozen U.S. states passed laws that prohibit healthcare employers from mandating overtime for nurses. Using a nationwide panel data set from 2004 to 2012, we find that these mandatory overtime laws reduced the service quality of nursing homes, as measured by an increase in deficiency citations. This outcome can be explained by two undesirable changes in the staffing hours of registered nurses: decreased hours of permanent nurses and increased hours of contract nurses per resident day.

n MB58 West Bldg 101C Improving Healthcare Accessibility and Performance Sponsored: Health Applications Sponsored Session Chair: Yong-Hong Kuo, The University of Hong Kong, Pokfulam Road, Hong Kong 1 - Location Model with Patient Choices and Care Facility Attractiveness Taesik Lee, KAIST, 291 Daehak-ro, Yuseong-gu, Daejeon, 305-701, Korea, Republic of, Kyosang Hwang Addressing health care needs of medically underserved areas is an important public health task. One solution is to establish care capacities in the underserved areas, and naturally this involves a location decision. To maximize the effectiveness of the investment, it is important for a location model to consider patients’ choice behavior to reflect future, expected use of new capacities. We present a location model with patient choice behavior in which care facilities’ attributes are integrated as decision variables. In addition to location decisions, the model determines the attributes of new facilities, providing additional levers to improve accessibility in the medically underserved areas. 2 - Physician Staffing and Shift Scheduling at Emergency Departments with Time Varying Productivity Alireza Sabouri, Haskayne School of Business, University of Calgary, 2500 University Dr. NW, SH132, Calgary, AB, T2N 1N4, Canada, Negar Ganjouhaghighi, Marco Bijvank Productivity of servers in service systems can decrease during a shift. This variable productivity alongside the stochastic nature of number of arrivals to the system, create a mismatch between demand and the number of customers that can be served by the scheduled servers during a particular period of time. In this study, we propose a two-step stochastic formulation for the staffing and shift scheduling problem with the objective of minimizing this mismatch. Numerical experiments are performed with data from a Canadian emergency department. The results show that the schedule generated by our formulation results in lower mismatch. Yong-Hong Kuo, Assistant Professor, The University of Hong Kong, Dept. of Industrial and Manu. Sys. Eng., The University of Hong Kong, Pokfulam Road, Hong Kong, Janny M. Y. Leung, Colin Graham This talk presents our work which uses simulation to analyze patient flows in a hospital emergency department (ED) in Hong Kong. This simulation approach provides a tool for the operations manager in the ED to assess the impact of changes in the system on the daily operations. We will discuss how simulation can be integrated into an optimization algorithm to aid decision-making. We will also present insights into managing ED operations derived from the simulation experiments. 4 - Appointment Scheduling with No-shows and Multiple Types of Patients Jingyao Huang, PhD Student, The University of Texas at Austin, 2110 Speedway,Austin, TX 78705, Austin, TX, 78705, United States, Douglas Morrice, Diwakar Gupta We consider an appointment scheduling problem with heterogeneous patients under no-shows. We have two types of patients. One is the regular patient who visits the clinic in person and has no-show behavior. The other is Virtual Medicine (VM) patient who receives service via e-visit and has a time window, within which they’ll be called and served. We first study a static model of assigning VM patients given regular patients’ schedule to maximize the expected profit. We partially characterize the optimal schedule and then extend the model to the sequential scheduling problem with random service time. Finally, we propose a heuristic to solve the problem quickly and effectively. We then evaluate the schedules by a simulation model. 3 - Simulation Analytics of Hospital Emergency Department Operations

n MB59 West Bldg 102A Incentives for Optimal Treatment Sponsored: Health Applications Sponsored Session

Chair: Greg Zaric, Ivey Business School, London, ON, N6G 0N1, Canada Co-Chair: Greg Critchley, Ivey Business School, London, ON, N6L0B2, Canada 1 - Incentives for Optimal Coordination in Multi-payer Health Care Systems

169