Informs Annual Meeting 2017

MA12

INFORMS Houston – 2017

MA12

MA13

332B Resource Allocation and Matching Sponsored: Manufacturing & Service Oper Mgmt,

332C Health Care, Processes Contributed Session

Service Operations Sponsored Session Chair: Weihua Zhang, University of British Columbia, Vancouver, BC, V6T 2G9, Canada, icystrawberry@hotmail.com Co-Chair: Yichuan Ding, University of British Columbia, Vancouver, BC, V6T 1Z2, Canada, daniel.ding@sauder.ubc.ca 1 - Advance Service Reservations with Heterogeneous Customers Xinshang Wang, Columbia University, New York, NY, United States, xw2230@columbia.edu, Clifford Stein, Van-Anh Truong We study a model of resource allocation in which a finite number of resources must be assigned in an online manner to a heterogeneous stream of customers. The customers arrive randomly over time according to known stochastic processes. Each customer requires a specific amount of capacity for each of the resources. The system must find a feasible assignment of each customer to a resource or must reject the customer. The aim is to maximize the total expected capacity utilization of all the resources. We present online algorithms with bounded competitive ratios. Our algorithms perform extremely well compared to common heuristics as demonstrated on a real data set from a large hospital system in New York City. 2 - Ohm’s Law Approximation for FCFS Stochastic Matching Edward H.Kaplan, Beach Professor of Operations Research, Yale University, Evans Hall, 165 Whitney Avenue, New Haven, CT, 06511, United States, edward.kaplan@yale.edu, Mohammad Fazel-Zarandi The FCFS stochastic matching model developed from queueing problems addressed by Kaplan (1984) for public housing assignments. The goal of this model is to determine the matching rates between eligible customer and server types. This model was solved in a beautiful paper by Adan and Weiss (2012), but the resulting equation for the matching rates is quite complicated, involving the sum of terms over all permutations of the server types. Here we develop approximations for the matching rates based on Ohm’s Law. As our approximations only require the solution of a system of linear equations along with the identification of a small number of terms, they provide a tractable alternative to the exact solution. 3 - A Novel KPD Mechanism to Increase Transplants when Some Candidates have Multiple Willing Donors Tuomas W. Sandholm, Carnegie Mellon University, Gates Center for Computer Science, Pittsburgh, PA, 15213, United States, Most current kidney exchanges allow candidates to list multiple willing donors, but only one will donate if the candidate is matched. There are cases where multiple donors would be willing to donate if that resulted in their intended candidate being transplanted. We conducted a dynamic simulation (using real UNOS kidney exchange data) to study the effects of having two donors donate in such cases. The new approach yielded more than a 10% gain in the number of priority-weighted transplants. This also means shorter wait times. Future research includes simulating other variants, studying the candidates’ and donors’ incentives, and considering the ethics. 4 - A Centralized Allocation Mechanism for Public Housing Weihua Zhang, University of British Columbia, Mailbox 66 0, 6335 Thunderbird Crescent, Vancouver, BC, V6T.2G9, Canada, weihua.zhang@sauder.ubc.ca, Yichuan Ding, Daniel Granot, Mahesh Nagarajan In 2014, US programs of federal rental assistance helped nearly five million families. We model the affordable housing allocation problem as a centralized decision problem. The HA strives to accommodate move-in requests as well as requests of transfer between housing types. We assume that households list the housing types that they would accept without specifying their complete preference order. We further assume that the housing units of each type are homogeneous. For minimizing vacancy rate and maximizing portability (the responsiveness of the HA to households’ accommodation requests), we formulate the problems as network flow problems and characterize the optimal solutions. sandholm@cs.cmu.edu, Gabriele Farina, John Dickerson, Ruthanne Leishman, Darren Stewart, Richard Formica, Carrie Thiessen, Sanjay Kulkarni

Chair: Haoqiang Jiang, Florida International University, 800 SW 104th Ct A-105, Miami, FL, 33174, United States, jhq820422@hotmail.com 1 - The Effect of Failure on Individual Performance Over Time Emmanouil Avgerinos, Assistant Professor, IE Business School, c/ María de Molina, 13, CIF.- G81711459, Madrid, 28006, Spain, emmanouil.avgerinos@ie.edu, Bilal Gokpinar Learning from past failures constitutes a key factor in modern organizations. While the phenomenon has been widely studied, the effect of failure over time has received little attention. More importantly, little is known about what management should do right after a failure is experienced by its employees in terms of team and task allocation. Using a unique dataset of 4,306 cardiac surgery operations from the cardiac unit of a private hospital in Europe we investigate the effect of failure over time and propose new ways for managers to help their employees recover from their recent failures. 2 - A Bayesian Approach for Estimating Out-of-network Referral Patterns in Accountable Care Organizations Iulian Ilies, Northeastern University, 360 Huntington Avenue, 1200 - 177, Boston, MA, 02115, United States, i.ilies@neu.edu, Parth R. Vadera, James C. Benneyan Outside utilization (patients receiving care from out-of-network providers) considerably limits the ability of accountable care organizations to coordinate care, ensure quality standards are followed, and reduce costs. We developed a Bayesian method for inference of referral links between care providers from historical insurance claims records, and evaluated this approach on synthetic data generated from provider networks of different sizes and topologies. Our approach was able to reconstruct the network structure with high accuracy, thereby providing an avenue to adjust referral patterns and reduce outside utilization. 3 - Implication of Codified Knowledge Sharing on Operational Failures Focusing on codified knowledge sharing among healthcare personnel in the form of written guidance documents, we examined the role of codified knowledge on operational failures in healthcare. We used a unique dataset from hospitals in England and employed text-mining techniques to investigate the impact of documents on operational failures. 4 - Development and Evaluation of Context-based Work Planning Policies for Inpatient Care Lucy Aragon, Student, Wichita State University, 1845 Fairmount Street, Wichita, KS, 67260, United States, lgaragon@shockers.wichita.edu, Lucy Aragon, Student, Pontificia Unviersidad Catolica del Peru, Lima, Peru, lgaragon@shockers.wichita.edu, Laila Cure The short-term inpatient care work planning problem consists of assigning tasks to hourly rounds. Due to the fast pace of healthcare systems and the cognitive limitations of humans, work planning strategies are often static and may be inadequate on particular workdays. This research studies work planning policies that can be applied by healthcare workers using limited computational resources. The problem was formulated as a variation of the 0-1 generalized assignment problem with nonlinear capacity constraints. Current, optimal, and practical work planning strategies were evaluated in terms of quality and added workload to the provider. 5 - Integrated Forecasting Simulation Approach for Call Center Workforce Management Bilal Gokpinar, UCL, Management Science/Innovation, Gower Street, London, WC1E 6BT, United Kingdom, b.gokpinar@ucl.ac.uk, Mecit Can Emre Simsekler Call center operations management has been a challenging task due to rising cost of operations. Major source of loss in the call center expenses is associated with poor staffing, such as over-staffing or under-staffing. The call center staffing is challenging because of a random and stochastic behavior of the demand, that is, the inbound call volume. This research aims to create a practical, robust, and flexible decision making framework for more efficient and effective planning of the staffing and scheduling in contact centers. This framework is based on demand forecasting and a simulation model of the call center operations combined with the forecasting data. Mohammadsadegh Mikaeili, Binghamton University, 1815 Kater Street, Philadelphia, PA, 19146, United States, mmikaei1@binghamton.edu

136