2016 INFORMS Annual Meeting Program

WA83

INFORMS Nashville – 2016

WA81 Broadway F- Omni Health Care, Strategies I Contributed Session Chair: Luv Sharma, Ohio State University, 601 Tuscarawas Court, Columbus, OH, 43210, United States, sharma.154@osu.edu 1 - Coordinated Scheduling Policies For Improving Patient Access To Surgical Services Mustafa Y Sir, Mayo Clinic, 200 First Street SW, Rochester, MN, 55905, United States, sir.mustafa@mayo.edu, Maria Gabriela Martinez, Todd Huschka, Kalyan Pasupathy Delayed access to receive treatment negatively affects patient satisfaction and health outcomes. This study presents scheduling policies that integrate patient flow in an elective surgical department in order to match capacity to patient needs. A data-driven model is formulated to determine appropriate scheduling time bounds for a fair distribution of surgical load among surgeons, considering medical need of patients. A simulation model is implemented to evaluate the performance of the proposed policies. 2 - Assessing Decisions In Medical Referral Networks From Empirical Data Michael Pavlin, Wilfrid Laurier University, School of Business and Economics, 75 University Avenue West, Waterloo, ON, N2L 3C5, Canada, mpavlin@wlu.ca, Mojtaba Araghi Informal referral networks are central to the allocation of medical resources in many healthcare systems. In this paper we assess decisions in a cataract surgery referral network. The system is modeled as a bipartite queueing network and empirical techniques are developed to estimate decision making parameters from aggregate data. 3 - A System Dynamics Model To Investigate The Impacts Of Non-invasive Sensor Based Interventions Mehmet Serdar Kilinc, Pennsylvania State University, 310 Leonhard Building, Department of Industrial Engineering, University Park, PA, 16802, United States, serdarmehmet@gmail.com, Jose Angel Castro, Linlin Ma, Harriet Black Nembhard Non-invasive sensor based interventions promise opportunities to reduce the burden of Parkinson’s disease (PD) on our healthcare system. These interventions can be implemented for various purposes such as early detection, remote monitoring of medication adherence, physical rehabilitation, and fall detection. To date, however, the US healthcare system has been slow to adopt these interventions. In this study, we use a system dynamics model to examine the long-term macro-level impacts of non-invasive sensor based interventions on the healthcare system. The model addresses both healthcare demand and supply by considering the prevalence and progress of PD with and without these interventions. 4 - Does The Office Of Patient Experience Matter In Improving Delivery Of Care? An Econometricstudy Of US Hospitals Luv Sharma, Ohio State University, 601 Tuscarawas Court, Columbus, OH, 43210, United States, sharma.154@osu.edu, Aravind Chandrasekaran We assemble a unique dataset regarding the presence of office of patient experience for 3250 US hospitals. We study the impact of this office on patient satisfaction. Results indicate that effectiveness of this office depends on the background of the chief experience officer, clinical complexity and its year of origin.

2 - Network Design Problem With Relays

Baris Yildiz, Assistant Professor, Koc University, Rumeli Feneri Yolu, Sariyer, Istanbul, Turkey, baris.yildiz@bilkent.edu.tr, Oya Ekin Karasan, Hande Yaman We study the network design problem with relays and present a multi-commodity flow formulation and a branch-and-price algorithm to solve it. Motivated by the practical applications we investigate the special case where each demand has a common designated source. In this special case, we can show that there exists an optimal design that is a tree. Using this fact, we replace the multi-commodity flow formulation with a tree formulation enhanced with Steiner cuts. Employing a branch-and-price-and-cut schema on this formulation, we are able to further extend computational efficiency. 3 - A Sampling Strategy For Estimating Features Of Large Networks Jingjing Zou, Columbia University, 434 W.120th St, Apt 3J, New York, NY, 10027, United States, jz2335@columbia.edu, Richard Davis, Gennady Samorodnitsky, Zhi-Li Zhang We propose a sampling procedure with the goal of estimating certain population features of the entire network. Such features might include tail behavior of the in- and out-degree distributions. Our procedure is based on selecting some initial nodes and then following the path of linked nodes in a structured fashion. Targeted nodes with desired features will have a larger probability of being retained. In order to construct unbiased or nearly unbiased estimates of the quantities of interest, weights associated with the sampled nodes must be calculated. We will illustrate this procedure and compare it with multiple random walks on datasets including webpage network and Google+ social network data. 4 - Games On Influence Networks: Equilibria, Free Riding And Dynamics Zhengyuan Zhou, Stanford University, 160 Comstock Circle, Unit 106002, Stanford, CA, 94305, United States, zyzhou@stanford.edu, Nicholas Bambos, Peter Glynn An influence network consists of a set of interacting agents, each of whose actions produces effects on his neighbors’ actions. In general, the effects can be arbitrary, inhomogeneous functions of the neighbors’ joint action. Such effects capture the commonality of a variety of networks in economics and engineering. Therefore, the study of strategic interactions among agents in an influence network can be of great applicability. We formulate a simple game-theoretical model of influence networks that investigates strategic interactions among agents in light of such influence. We study the equilibria properties, the resulting free-riding phenomenon and the dynamics for reaching an equilibrium. Chair: Stewart Liu, PhD Candidate, Univeristy of California - Berkeley, 4141 Etcheverry Hall, MC 1777, Berkeley, CA, 94720, United States, stewart_liu@berkeley.edu 1 - A Newsvendor Problem With Multiple Unreliable Suppliers Roshanak Mohammadivojdan, PhD Student, University of Florida, 303 Weil Hall, P.O. Box 116595, Gainesville, FL, 32611, United States, rmohammadivojdan@ufl.edu, Yasemin Merzifonluoglu, Joseph Geunes We consider a single period (newsvendor) inventory planning problem in which the newsvendor must determine how much to order from each of a number of suppliers, and where each supplier may default on delivery with a certain probability. Our goal is to minimize total expected cost, including total ordering costs and overstocking and understocking costs. We characterize key structural properties of optimal solutions and provide algorithms for solving problems in this class. 2 - Topological Network Design Of Closed Finite Capacity Supply Chain Networks Laoucine Kerbache, Professor, HEC Paris School of Management, Ops Mgmt & Information Systems, 1 Rue de La Liberation Jouy-En-Josas, 78351, France, kerbache@hec.fr, James Macgregor Smith We wish to examine the layout, location, and general topological arrangement of queues in a closed queueing network environment for supply chains. Since our focus is on manufacturing environments, then maximizing throughput is a worthy performance measure objective. We are given a network topology G(V,E) with a finite set of nodes and edges and we wish to assign the queues to the nodes such that the maximum throughput is achieved. We examine the performance of the system in a closed queueing network environment. This is a nonlinear continuous optimization problem with implicit integer variables. We show that decentralization can have a major impact on the throughput of the supply chain. WA83 Broadway H- Omni Supply Chain Optimization I Contributed Session

WA82 Broadway G- Omni Networks and Graphs I Contributed Session

Chair: Zhengyuan Zhou, Stanford University, 160 Comstock Circle - Unit 106002, Stanford, CA, 94305, United States, zyzhou@stanford.edu 1 - Facets And Valid Inequalities For The Pathwidth Problem Tom Rihm, University of Bern, Schuetzenmattstrasse 14, Bern, Switzerland, tom.rihm@pqm.unibe.ch, Arie Koster The pathwidth specifies the similarity between a given graph and a path, and is relevant for many algorithms in real-world applications. In general, the problem of determining the pathwidth of a given graph is NP-complete. We formulate this problem as an integer linear program. Furthermore, we provide classes of valid inequalities to tighten the linear programming relaxation, and we identify conditions under which these inequalities are facet-defining. Our computational results indicate that these inequalities improve the performance.

391