Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MB72

4 - A Center Cut Algorithm for Quickly Obtaining Feasible Solutions and Solving Convex Mixed Integer Nonlinear Programs David E. Bernal, Carnegie Mellon University, 5000 Forbes Avenue, Doherty Hall office 3116, Pittsburgh, PA, 15213, United States, Jan P. Kronqvist, Andreas Lundell, Tapio Westerlund, Ignacio E. Grossmann We present a center-cut algorithm for convex mixed-integer nonlinear programs that can be used as a primal heuristic or as a deterministic solution strategy. The trial solutions are chosen at the analytical center of the linear approximation of the nonlinear constraints subject to the original integrality constraints, with the idea that this trial solution is likely to be feasible. At this solution, the linear approximation is refined. The procedure is repeated iteratively, leading to convergence to a feasible solution. Proofs for the finite convergence of the algorithm are provided, and computational results show that the algorithm quickly finds feasible solutions to convex MINLPs. Joint Session Wagner/Practice: Daniel H. Wagner Prize for Excellence in Operations Research Practice Emerging Topic: Daniel H. Wagner Competition Emerging Topic Session Chair: Patricia Neri, SAS Institute, Inc., 104 Grandtree Ct., Cary, NC, 27519, United States 1 - Primal-Dual Algorithms for Order Fulfillment at Urban Outfitters, Inc. Vivek Farias, MIT, 100 Main Street, Cambridge, MA, 02142, United States, John M. Andrews, Aryan Khojandi, Chad Yan We formulate the omni-channel fulfillment problem as an online optimization problem. We propose a novel algorithm for this problem based on the primal-dual schema. Our algorithm is robust: it does not require explicit demand forecasts. This is an important practical advantage in the apparel-retail setting where demand is volatile and unpredictable. We provide a performance analysis establishing that our algorithm admits optimal performance guarantees in the face of adversarial demand. We describe a large-scale implementation of our algorithm at Urban Outfitters, Inc. This implementation processes on average eighteen thousand customer orders a day, and as many as one hundred thousand orders on peak demand days. We estimate conservatively that the system has saved at least six million dollars annually relative to an incumbent industry standard fulfillment optimization implementation. This saving is achieved through optimal order- fulfillment decisions that simultaneously increase turn and lower shipping costs. 2 - Centralized Admissions for Engineering Colleges in India Yash Kanoria, Columbia Business School, 404 Uris Hall, New York, NY, 10027, United States, Surender Baswana, Partha Pratim Chakrabarti, Sharat Chandran, Utkarsh Patange We designed and implemented a new joint seat allocation process for over 500 programs across 80 engineering colleges in India, including the Indian Institutes of Technology (IITs). Our Deferred Acceptance-based process overcomes a number of challenges. The joint seat allocation process has been successfully running since 2015, with a 70% reduction in vacancies at the IITs. Particular concerns have been (i) allocative efficiency and reducing vacancies due to students accepting a seat but then withdrawing or not reporting for classes, (ii) transparency, integrity and fairness, (iii) logistical simplicity for both candidates and institutes, and (iv) the incentive structure for candidates. n MB73 West Bldg 211B Innovations in Healthcare Delivery Sponsored: Health Applications Sponsored Session Chair: Alex Mills, Kelley School of Business, Indiana University Bloomington, Bloomington, IN, 47405, United Statesu 1 - Healthcare Payment Model Impact on Hospital Readmissions Jon M. Stauffer, Assistant Professor, Texas A&M University, 4217 TAMU, College Station, TX, 77843, United States, Jonathan Helm, Kurt M. Bretthauer We examine how pay-for-performance (P4P) reimbursement plans, such as bundled payments and the Hospital Readmission Reduction Program (HRRP) impact the motivation for providers to reduce readmissions. Results show that P4P plans do motivate extra readmission reduction effort, but that misalignments can occur between the provider’s efforts and the minimum total system cost effort. We also find there is only a small window when the bundled payment price is large enough to motivate all providers to reduce readmissions, but not so n MB72 West Bldg 211A

large as to over-motivate a smaller provider to perform effort exceeding the minimum total system cost effort. 2 - Managing Patient Panels with Nonphysician Providers Hessam Bavafa, Wisconsin School of Business, 4284C Grainger Hall, 975 University Avenue, Madison, WI, 53706, United States, Sergei Savin, Christian Terwiesch In recent years, the drive to contain health care costs has increased scrutiny of the traditional mode of delivering primary care where a patient is treated only by his primary care physician. In particular, greater reliance on non-physician providers has been suggested as a lower-cost alternative to the traditional set-up. We study the overall impact of non-physician providers on the physician’s expected daily compensation and patients’ expected health. 3 - The Business of Healthcare: Physician Integration in Bundled Payments Turgay Ayer, Georgia Institute of Technology, School of Industrial and Systems Engineering, Groseclose 417, Atlanta, GA, 30332, United States, Jan Vlachy, Mehmet U.S. Ayvaci Under the prevailing fee-for-service payments, incentives of hospitals and physicians are misaligned, leading to large inefficiencies. Bundled payments unify payments to the hospital and physicians and are expected to encourage care coordination and reduce costs. However, as hospitals differ in their relationships with physicians in influencing care level of physician integration, it remains unclear what spectrum of physician integration will facilitate bundling. We study (1) the impact of the level of integration between the hospital and physicians in the uptake of bundled payments, (2) the consequences of bundling with respect to overall care quality and costs/savings. 4 - Estimated Time to Discharge Building the ETA for Hospitals Jean Pauphilet, MIT, Cambridge, MA, 02139, United States, Dimitris Bertsimas Number of beds, whose availability is driven by patient discharges, is one of the most critical resources in any hospital. As a result, accurate prediction of time-to- discharge could significantly improve daily operations. In this work, we combine advanced machine learning methods with expertise and insights from medical staff, social workers and operations to compute personalized Time-to-Discharge estimates for each individual inpatient. To that end, we also build a predictive model to anticipate discharge destination as well. Joint work with BIDMC, Boston, MA. n MB74 West Bldg 212A Exact Approaches and Approximations in Multiobjective Optimization Sponsored: Multiple Criteria Decision Making Sponsored Session Chair: Serpil Sayin, Koc University, Istanbul, 34450, Turkey 1 - An Algorithm for Solving Bi-objective Integer Programming Problems Ozlem Karsu, Bilkent University, Bilkent University, Ankara, 06800, Turkey, Firdevs Ulus, Saliha Dogan We propose an objective-space based algorithm to solve bi-objective integer programming problems, where we assume that the set of all nondominated points are included within a bounded set. In each iteration of the algorithm a single Benson-type (weighted Chebyshev) scalarization is solved. The computational experiments demonstrate the satisfactory performance of the algorithm. 2 - Tractability of Convex Vector Optimization Problems in the Sense of Polyhedral Approximations Firdevs Ulus, Bilkent University, Industrial Engineering Department, Bilkent, Ankara, 06800, Turkey A recent solution concept for convex vector optimization problems (CVOPs) which is motivated from a set optimization point of view consists of finitely many efficient solutions that generate polyhedral inner and outer approximations to the Pareto frontier. A CVOP with compact feasible region is known to be bounded and there exists a solution of this sense to it. However, it is not known if it is possible to generate polyhedral inner and outer approximations to the Pareto frontier of a CVOP if the feasible region is not compact. This study shows that not all CVOPs are tractable in that sense and gives a characterization of tractable problems in terms of the well known weighted sum scalarization problems.

174