2016 INFORMS Annual Meeting Program

WC32

INFORMS Nashville – 2016

WC30 202B-MCC Data-Driven Models in Healthcare Sponsored: Manufacturing & Service Oper Mgmt, Healthcare Operations Sponsored Session Chair: Yichuan Ding, University of British Columbia, 2053 Main Mall, Sauder School of Business, Vancouver, BC, V6T1Z2, Canada, daniel.ding@sauder.ubc.ca 1 - Bundled Payments And Value Based Delivery Jillian B Jaeker, Boston University, jjaeker@bu.edu This study aims to understand the care pathways, both inpatient and outpatient (i.e. end-to-end), for congestive heart failure (CHF) patients, and how these pathways impact the costs and quality of care. Specifically, our objective is to identify pathways that are associated with lower or higher than normal readmission rates, and the associated financial costs of these pathways. Moreover, we work with hospitals that are participating in the Bundled Payments for Care Improvement (BPCI) initiative and explore if any recent interventions on behalf of these hospitals affects the probability of 30 day readmission and total episode costs. 2 - Dynamic Allocation Of Vaccine Stocks For Pandemic Influenza Mitigation Ozden O Dalgic, University of Waterloo, Waterloo, ON, Canada, oodalgic@uwaterloo.ca Vaccination is our best response against influenza pandemics. However, finding effective vaccine allocation strategies with current modeling technics can be challenging. In this study, we propose a hybrid approach combining simulation and analytical modeling. We employ a model based on a chain-binomial transition scheme calibrated to an existing agent-based simulation model. The proposed approach efficiently evaluates the performance of a given vaccine allocation strategy by reducing the number of random variates. Numerical experiments show that the proposed approach results in effective dynamic vaccine allocation strategies. 3 - Designing Personalized Anticoagulation Therapy Rouba Ibrahim, University College London, rouba.e.ibrahim@gmail.com, Vedat Verter, Beste Kucukyazici, Michel Gendreau, Mark Blostein We develop an analytical framework for personalizing the anticoagulation therapy of patients who are taking warfarin, and present results from a case study using data collected at the anticoagulation clinic of the Jewish General Hospital in Montreal. 4 - Do Patients From Rural Areas Get Proper Referral For Surgical Care Yichuan Ding, University of British Columbia, 2053 Main Mall, Vancouver, BC, V6T1Z2, Canada, daniel.ding@sauder.ubc.ca We examined surgical records throughout 98 hospitals in British Columbia during year 1995-2004, and had two interesting observations: (1) hospitals in rural areas are correlated with smaller likelihood of post-surgical complications, possibly because those hospitals have less risky case mix groups; (2) patients from rural areas, however, are correlated with lower risk of post-surgical complications. We conduct further investigation and find that the reason for (2) to happen might be the inefficient communication between the referral hospital and the one that the surgery takes place.

that under the standard exogenous abandonment model, and to its empirical counterpart in a real system. 2 - Learning Quality Through Service Outcomes Senthil Veeraraghavan, The Wharton School, University of Pennsylvania, senthilv@wharton.upenn.edu, Laurens G Debo We study a new firm whose service value is unknown to arriving customers. Service outcomes are random depending on the quality of the service provider. Customers decide whether patronize the service based on the limited service outcomes/reviews that they observe. We consider how service policies influence consumer learning and social welfare. 3 - Design Of Discretionary Service Lines: An Operational Driver Of Variety Laurens Debo, Tuck School of Business, Dartmouth College, 100 For discretionary services, the longer the service time, the more value is created for the customer. In the presence of variability, longer service times also create more congestion. Hence, a service firm needs to trade off congestion costs with value creation in service line design. We find that it is optimal to offer a service line with multiple varieties (that differ in duration and price), even when customers are homogeneous. 4 - Linking Customer Behavior And Delay Announcements: Are Customers Really Rational? Tuck Drive, Hanover, NH, 03755, United States, Laurens.g.Debo@tuck.dartmouth.edu, Cuihong Li We empirically explore customer abandonment behavior in the presence of delay information using data from a call center. Previous work has assumed that customers are at least partially rational in responding to announcements. In contrast, we relax all rationality assumptions. Our findings indicate that customers exhibit loss aversion behavior. In addition, customers may update their announcement-induced reference point as they hear subsequent announcements. Our results also indicate that customers may fall for the sunk cost fallacy while waiting in the queue. We show the impact of these effects on staffing decisions using a classic staffing model. Risk Analysis III Contributed Session Chair: Dexiang Wu, Stockholm University, Stockholm Business School, Stockholm, 106 91, Sweden, dexiang.wu@sbs.su.se 1 - Modeling Behavior Of Attackers Against The Uncertainty Of Cascading Failure Sinan Tas, Assistant Professor, University of Wisconsin-Platteville, 1 University Plaza, Platteville, WI, 53818, United States, tass@uwplatt.edu Cascading failure is a common phenomenon in capacity-constrained networks such as power grids. What if attackers consider manipulating the additional impact of cascading failure to enhance the damage to the network? We analyze various attacker types and corresponding defensive investments when cascading failure is modeled stochastically in a game-theoretic setting. 2 - Analysis Of Intraday Data Effects On Two-stage Risk-averse Portfolio Optimization Sitki Gulten, Stockton University, School of Business, 101 Vera King Farris Drive, Galloway, NJ, 08205, United States, sitki.gulten@stockton.edu This study examines the application of risk-averse optimization techniques to daily portfolio management. First, I develop efficient clustering methods for scenario tree construction. Then, I construct a two-stage stochastic programming problem with conditional measures of risk, which is used to re-balance the portfolio on a rolling horizon basis, with transaction costs included. Finally, I present an extensive simulation study on both interday and high-frequency intraday real-world data of the methodology. 3 - Exploring Multi-stage Recovery Resilience Daniel Hernando Romero, University of South Florida, 4411 Shady Terrace Ln, Unit A, Apt 212, Tampa, FL, 33613, United States, danielromero@mail.usf.edu, Alex Savachkin, Alvaro Sierra, Weimar Ardila Multi-stage recovery models enable resilience measurement in the scenarios where the recovery time is long, so it is estimated en months or years. High magnitude disruptive events severely affect communities and generate long lasting consequences. These scenarios require model flexibility to capture different recovery rates and transitions between recovery stages. Eric Webb, Kelley School of Business, Indiana University, Bloomington, IN, United States, ermwebb@indiana.edu, Qiuping Yu, Kurt M Bretthauer WC32 203A-MCC

WC31 202C-MCC Consumer Behavior in Services Sponsored: Manufacturing & Service Oper Mgmt,

Service Operations Sponsored Session

Chair: Qiuping Yu, Assistant Professor, Kelley School of Business, Indiana University, 1275 E 10th St, Bloomington, IN, 47401, United States, qiupyu@indiana.edu 1 - Rational Abandonment From Observable Priority Queues Philipp Afèche, Rotman School of Management, University of Toronto, Toronto, ON, Canada, afeche@rotman.utoronto.ca, Vahid Sarhangian The literature on customer behavior in queueing systems largely focuses on customers’ joining decisions and ignores their subsequent abandonment decisions. Such abandonment behavior is important in priority queues, which are prevalent in practice. We characterize the equilibrium joining and abandonment behavior of utility-maximizing customers in an observable priority queue. We then discuss how the abandonment process in our equilibrium model compares to

437