2016 INFORMS Annual Meeting Program

WA78

INFORMS Nashville – 2016

WA78 Legends F- Omni Opt, Metaheuristics I Contributed Session

combined SAA approach. Results also indicate that, strengthened formulation which is typically much faster to solve chance constrained MILP, can also be outperformed by the enhanced SAA approach for larger instances.

WA80 Broadway E- Omni Health Care, Public I Contributed Session Chair: Amit K Bardhan, Professor, University of Delhi, Delhi, 110007, India, amit-bardhan@fms.edu 1 - Application Of System Dynamics To Private Sector Versus Medicare Cost Projections Michael P D’Itri, Associate Dean and Professor, Dalton State College, 650 North College Drive, Dalton, GA, 30720, United States, mditri@daltonstate.edu, Robert Culp, Jon Littlefield A source of contention over the Affordable Care Act results from analyses based on segmentation of the population studied into public (Medicare) versus private healthcare providers. This research controls for this effect by employing a systems dynamic method to model private sector and Medicare costs over time. Age data from the 2000 United States Census is represented using a distributed delay with net annual changes in population calibrated to match the 2010 census data. Average costs per person in each age group are calculated and used to make future cost projections. 2 - Identifying The Impact Of Outdated Drug Limit Library Usage By Smart Infusion Pump Logs Kang-Yu Hsu, PhD Student, Purdue University, Gerald D and Edna E Mann Hall, 203 S. Martin Jischke Drive, Lafayette, IN, 47907, United States, hsu66@purdue.edu, Poching DeLaurentis, Yuval Bitan, Daniel Degnan, Yuehwern Yih Drug Limit Library (DLL) in infusion pumps may not be updated efficiently. The inconsistency between DLL in pumps and the most up-to-date DLL may put patient safety at risk. In this study, we quantify the impact of adopting out-of-date DLL through investigating smart infusion pump logs, and examine infusions which potentially jeopardized patient safety with outdated DLL usage. 3 - A Game Theoretic Approach To Pediatric Vaccine Pricing Banafsheh Behzad, Assistant Professor, California State University, Long Beach, 1250 Bellflower Blvd, MS 8506, Long Beach, CA, 91101, United States, banafsheh.behzad@csulb.edu, Sheldon H Jacobson Pricing strategies in the US pediatric vaccines market are studied using a Bertrand-Edgeworth-Chamberlin price game. The game analyzes the competition between asymmetric, capacity-constrained manufacturers producing differentiated products in a market with linear demand. The results indicate that the pure strategy equilibrium exists if the production capacity of a manufacturer is at their extreme. In a duopoly setting, the distribution functions of the mixed strategy equilibrium for manufacturers are provided. The proposed game is applied to the US pediatric vaccine market, where competing vaccines are differentiated based upon the number of reported medical adverse events. 4 - Storage And Transport Considerations In Designing The WHO-EPI Vaccine Distribution Network Jayant Rajgopal, Professor, University of Pittsburgh, Department of Industrial Engineering, University of Pittsburgh, Pittsburgh, PA, 15261, United States, rajgopal@pitt.edu, Jung Lim, Bryan A Norman The WHO-EPI vaccine distribution chain is used to deliver vaccines for inoculating children against vaccine-preventable diseases. The structure of this chain is rigid and almost identical in most low and middle income countries, despite significant demographic, geographical and economic differences. We describe the problem of designing a network that is based on the parameters associated with a specific country and present an integer programming formulation that optimizes the structure for a given country while also determining the primary parameters of cold storage and transportation along with the design. Issues related to solving this model are also discussed. 5 - Healthcare Facility Location Model Based On Choice Behaviour Of Catchment Population Amit K Bardhan, Professor, University of Delhi, Delhi, 110007, India, amit-bardhan@fms.edu, Arun K Sharma, Kaushal Kumar Due to excessive demand and limited capacity, public healthcare delivery systems in developing countries operate under immense pressure. New facilities when established should not only be useful to the unserved, they should also reduce pressure on existing centers. Most healthcare facility location models focus on ease of access criteria like distance, travel-time, population density etc. In recent studies it has been reported that assurance and quality of care are also important while choosing healthcare facility. In this paper we propose a hierarchical facility location model that incorporates such behavioral choice criteria of target population.

Chair: Pakayse Koken, PhD Candidate, Binghamton University, 4400 Vestal Parkway East, Binghamton, NY, 13902, United States, pakoken@gmail.com 1 - Optimizing Communication In Parallel Algortihm Portfolios Andrii Berdnikov, Graduate Student, University of Tennessee, Knoxville, 525 John D. Tickle Building,851 Neyland Drive, Knoxville, TN, 37996-2315, United States, andrii@utk.edu We establish a Markov model of parallel algorithm portfolio performance that captures communication between individual algorithms. Based on the proposed model we investigate different probabilistic measures of efficiency and speed to evaluate performance of algorithm portfolios. These measures are used to optimize communication between individual algorithms in portfolio configuration. 2 - Metaheuristics For Dynamic Lot Sizing Problem With Returns And Hybrid Products Pakayse Koken, PhD Candidate, Binghamton University, 4400 Vestal Parkway East, Binghamton, NY, 13902, United States, pakoken@gmail.com For a hybrid system with manufacturing and remanufacturing, a variant of dynamic lot sizing problem is addressed. In the hybrid system, manufactured, remanufactured and hybrid products are produced. Hybrids are composed of approximately 90% new parts and 10% returns. The main objective of this study is to investigate the profitability conditions for producing hybrids. Therefore, a variant of dynamic lot sizing problem is formulated as a mixed-integer nonlinear programming (MINLP) problem. The performance of the system with hybrids is compared to the same system with no hybrids. Metaheuristic algorithms are used to find near optimal solutions to the MINLP problem. 3 - A Hybrid Genetic Algorithm For The Fixed Charge Transportation Problem With The Non-linear Unit Costs Kiseok Sung, Gangneung-Wonju National University, Gangneung, Korea, Republic of. Contact: sung@gwnu.ac.kr We present a hybrid method of the Genetic Algorithm for the Fixed Charge Transportation Problem with the Non-linear Unit Costs, mathematically formu- lated with the 0-1 mixed integer program with non-linear objective function and linear constraints. The GA is used in the upper level to optimize the connectivity of the transportation route between each supply and demand pair. The Continuous GA is used in the lower level to optimize the amount of transporta- tion between each supply and demand pair. In the upper and lower level of pro- cedure, the solutions are verified of the feasibility and modified if necessary to maintain the feasibility. Opt, Stochastic V Contributed Session 1 - A New Framework For Shortest Path Problem In Dynamic Network Yingying Kang, Senior Operations Research Developer, Southwest Airlines Co., Dallas, TX, United States, eing.008@gmail.com Shortest path algorithm has been well developed over years. The classic dijkstra’s algorithm provides exact solution with reasonable run time. However, it has limitation in solving complicated problems, especially when network is dense and changes dynamically. This presentation presents a new framework and enhanced searching procedure to raise the efficiency in dynamic network. This solution has been proved to improve the solution quality and efficiency significantly in large scaled dynamic network and provide a practical and effective solution to real time problems. 2 - An Enhanced Sample Average Approximation Technique To Solve A Two-stage Chance-constrained Optimization Problem Sudipta Chowdhury, PHD Student, Mississippi State University, 260 McCain Engineering Building, ISE Department, Starkville, MS, 39762, United States, sc2603@msstate.edu, Adindu Emelogu, Mohammad Marufuzzaman, Linkan Bian This study proposes an enhanced Sample Average Approximation (SAA) technique to solve a two-stage stochastic chance-constrained optimization problem. The problem is challenging to solve as the feasibility region defined by chance constraints is generally non-convex and hence requires multi-dimensional integration. The numerical studies show that for all sized instances the combined enhanced SAA approach gives faster and better quality solution than the WA79 Legends G- Omni

390