2015 Informs Annual Meeting

WA14

INFORMS Philadelphia – 2015

3 - Decomposition Algorithm for Two-stage Stochastic Integer Programs with Deviation Risk Measures Saravanan Venkatachalam, Texas A&M University, 3131 TAMU, College Station TX 77843,saravanan@tamu.edu, Lewis Ntaimo, Bernardo Pagnoncelli In this talk, we present a decomposition algorithm for two-stage stochastic integer programs (SIP) with absolute semi-deviation (ASD) risk measures. These types of problems are hard to decompose due to the lack of amenable problem structure. We propose a methodology that uses the information from expected excess risk measure to solve SIPs with ASD risk measures. Preliminary computational results will be presented. WA14 14-Franklin 4, Marriott Advances in Information Modeling Sponsor: Optimization/Optimization Under Uncertainty Sponsored Session Chair: Eugene Perevalov, Lehigh University, 200 W Packer Ave, Bethlehem, PA, United States of America, eup2@lehigh.edu 1 - Information Sources’ Knowledge Structure, Question/Answer Depth and Information Relevance Eugene Perevalov, Lehigh University, 200 W Packer Ave, Bethlehem, PA, United States of America, eup2@lehigh.edu, Xing Wang To optimize the process of information acquisition, a problem solving agent analyzes both the available information and the information relevant to the given problem, and, based on such analysis, selects the information source(s) that can help obtain the largest improvement of the solution quality. Currently, this is done on a largely ad hoc case-by-case basis. Our goal is to develop a general theory providing a regular way of performing such a selection. 2 - Towards the Full Information Theory: Knowledge Structure of Information Sources Xing Wang, Lehigh University, 200 W Packer Ave, Bethlehem, PA, United States of America, xiw313@lehigh.edu, Eugene Perevalov The classical Information Theory was able to properly describe the sheer quantity of information which allowed for optimization of information transmission leaving the process of information acquisition still in the “pre-theoretical” state relying on ad-hoc approaches. We initiate the study of the latter which makes it necessary to describe the ability of information sources to change the agent’s information state depending on the specific information requested. 3 - Using Network Diversity in Analytics and Visualization Seifu Chonde, Student, The Pennsylvania State University, 311 Leonhard Building, University Park, PA, 16802, United States of America, sjc294@psu.edu, Soundar Kumara In 2007, the Rao-Stirling diversity framework was introduced to capture changes in the variety, balance, and disparity among nodes in networks. We extended this diversity measure to include coherence, which captures network density. We show how our diversity measure improves approaches in clustering, link prediction, and tracking evolutionary networks. We use a case study in drug design with hundreds of thousands of journal articles to motivate the study of diversity.

and individual behaviors. 2 - Agent-Based Mediation Modeling of a Controlled Trial to Reduce Transmission of Resistant Organisms Sean Barnes, University of Maryland, 4352 Van Munching Hall, University of Maryland, College Park, MD, 20742, United States of America, sbarnes@rhsmith.umd.edu, Daniel Morgan, Lisa Pineles, Anthony Harris In 2012-2013, the University of Maryland School of Medicine led a 20-site trial to assess the benefits of universal gloves and gowns in reducing the transmission of multidrug-resistant organisms in acute-care hospitals. We develop an agent-based model to simulate transmission in an ICU, and calibrate this model to best match the outcomes from the 20 sites. Then, we validate the model against the study’s analysis, and perform a factorial design to quantify the direct benefit of this intervention. 3 - A Systems Approach to Reducing Central Line Associated Blood Stream Infections Eva Lee, Georgia Institute of Technology, eva.lee@gatech.edu The study aims to reduce central line associated bloodstream infections (CLABSI) at Grady Hospital by 20% from current levels. A systems model is designed that comprises the entire process of central line insertion and maintenance. Hospital data is analyzed to identify potential increased risk of CLABSI-related deaths. An second objective includes predicting death among CLASBI-infected patients while minimizing Type II errors; enabling providers to intervene early to reduce unnecessary deaths. 4 - Assessing CAUTI Prevention Interventions in Intensive Care Units Eduardo Pérez, Assistant Professor, Texas State University, 601 University Drive Ingram School of En, San Marcos, TX, 78666, United States of America, eduardopr@txstate.edu This research extend the current methods for addressing CAUTIs to a new level beyond the current state of practice by introducing a new simulation model for system assessment and determination of the clinical efficacy and effectiveness of CAUTI preventive interventions. The simulation model provide a platform where specific interventions can be computed and analyzed, giving an idea on possible expectations of the intervention before clinical research is conducted. Chair: Chuangyin Dang, Professor, City University of Hong Kong, Dept. of SEEM, 83 Tat Chee Avenue, Kowloon, Hong Kong - PRC, mecdang@cityu.edu.hk 1 - Royalty Licensing Mechanisms in a Leadership Structure Masashi Umezawa, Professor, Tokyo Univesity of Science, School of Management, 500 Shimokiyoku, Kuki, Saitama, 346-8512, Japan, m.umezawa@gmail.com This paper studies a patent licensing of a cost-reducing innovation which is developed by an internal patentee, one of the firms of the industry. Royalty licensing is mainly focused. Most patent licensing contracts observed empirically include either per-unit or ad valorem royalties. We consider that in a Stackelberg duopoly which types of royalty the patentee prefers. Moreover, we examine the effect of royalty licensing from the point of consumers and social welfare. 2 - Horizontal Product Differentiation in the Presence of Core Products Hiroki Kishihara, Keio University, Hiyoshi 3-14-1 Kohoku-ku, Yokohama, Japan, kishihara@z5.keio.jp, Nobuo Matsubayashi We study horizontal product differentiation between two firms by a game- theoretic approach. We suppose that to introduce their new products, firms must incur ``repositioning cost” which increases with the position difference from their core products exogenously given in a product space. We show that the presence of repositioning costs can have a significant impact on the equilibrium outcomes, which also depends on their cost structures. 3 - A General Method for Checking the Core Emptiness of Traveling Salesman Games Lei Sun, Development Specialist, Praxair Inc., 175 E. Park Dr., Tonawanda, NY, 14150, United States of America, Lei_Sun@Praxair.com, Mark Karwan We study the core of traveling salesman games from an optimization perspective and propose a new variant of the traveling salesman problem building a link between the two problems. An exact formula is found to measure the gap between the best achievable cost allocation and the core. With network optimization models and algorithms, we develop an implementable method to check whether an empty core exists in general for both symmetric and asymmetric traveling salesman games. WA16 16-Franklin 6, Marriott Game Theory II Contributed Session

WA15 15-Franklin 5, Marriott Monitoring and Prevention of Hospital Acquired Infections Sponsor: Optimization in Healthcare Sponsored Session

Chair: Eduardo Pérez, Assistant Professor, Texas State University, 601 University Drive Ingram School of Eng, San Marcos, TX, 78666, United States of America, eduardopr@txstate.edu 1 - Network Effects on Hand-hygiene Behavior Reidar Hagtvedt, University of Alberta, Edmonton, AB, Canada, hagtvedt@ualberta.ca, Brian Whiteside, Bruce Wright, Sarah Forgie Hand-hygiene in healthcare settings is a phenomenon that has been studied a great deal, both on the individual and on the group level, but to our knowledge detail on the group working as a network in this context has not been studied. We use sociometric badges to gather data from a large teaching hospital and estimate network statistics. We then model the relationship between the network

378