2016 INFORMS Annual Meeting Program

WA77

INFORMS Nashville – 2016

WA75 Legends C- Omni Health Care, Strategies III Contributed Session

2 - Fast Approximate Policies For Large Networks Ankur Mani, University of Minnesota, 808 Berry St. Apt. 410, St. Paul, MN, 55114, United States, amani@umn.edu Optimal policies for networks may be computationally hard and still may lead to suboptimal outcomes if the network information is noisy. We present simple heuristics with comparable performance. These policies are easy to compute and we provide guarantees for their performance. As an example we study price discrimination in networks and show that our heuristics give approximately optimal expected profits for large random networks. 3 - Multivariate Subexponential Distributions And Their Applications Julian Sun, Cornell, Ithaca, NY, 14850, United States, ys598@cornell.edu, Gennady Samorodnitsky We propose a new definition of a multivariate subexponential distribution. We compare this definition with the two existing notions of multivariate subexponentiality, and compute the asymptotic behavior of the ruin probability in the context of an insurance portfolio, when multivariate subexponentiality holds. Previously such results were available only in the case of multivariate regularly varying claims. WA77 Legends E- Omni Opt, Integer Programing V Contributed Session Chair: Victoire Denoyel, PhD Candidate, ESSEC Business School, Paris, France, victoire.denoyel@essec.edu 1 - Territory Design With Risk For A Micro Finance Institution Tahir Ekin, Assistant Professor of Quantitative Methods, Texas State University, 601 University Dr. Mccoy 411, San Marcos, TX, 78666, United States, t_e18@txstate.edu, Fabian Lopez Perez, Francis Mendez, Jesus Jimenez Micro finance institutions (MFIs) play an important role in emerging economies as part of programs that aim to reduce income inequality and poverty. This talk addresses a territory planning problem for a MFI. We propose a mixed integer programming model that lets the decision maker choose the location of the branches to be designated as territory centers and allocate the customers to these territory centers with respect to risk and planning criteria: the total workload, monetary amount of loans and profit allocation. In order to solve this model for the large size instances of the MFI, we utilize heuristics such as fixing variables, perturbation, and dynamic relocation of territory centers. 2 - Model Of Ambulance Deployment And Dispatch Arrangement And Simulation-based Assessment Of Mass Casualty Incident Yu-Ching Lee, Assistant Professor, National Tsing Hua University, Hsinchu, Taiwan, liuyentingthomas@gmail.com, Yen-Ting Liu, Albert Y. Chen, Yu-Shih Chen A major focus of emergency medical service (EMS) systems is to save lives, minimize response time and to increase the survival rate in both the cases of stochastic events and extreme events. The extreme events, such as natural disaster (earthquake) and man-made disaster (terrorist attacks), are usually not predictable and need multiple rounds of ambulance dispatch. This paper is committed to model the deployment and split dispatch of ambulances, and to generate numerical results to assess the overall services, hoping to be an optimization-model-aided support to the real world operations. 3 - Solving Utility-maximization Multinomial Choice Problems: When Is The First-choice Model A Good Approximation? Victoire Denoyel, PhD Candidate, ESSEC Business School, Paris, France, victoire.denoyel@essec.edu, Victoire Denoyel, PhD Candidate, Brooklyn College, Brooklyn, NY, NY, United States, victoire.denoyel@essec.edu, Laurent Alfandari, Aurelie Thiele For optimization problems with a utility maximization objective, it is common to model consumer behavior with MNL logit. This can lead to high fractional complexity when binary decision variables are involved. A first-choice or assignment model is computationally simpler although less close to reality. We design the first comparison of the two approaches in the context of optimization. Our main contribution is to quantify which probabilistic assumptions allow the use of the solution of the first-choice model as an approximation to the MNL logit model. Applications vary from policy to retail or facility location.

Chair: Neset Hikmet, Associate Professor, University of South Carolina, 1301 Gervais St, Suite 1010, Columbia, SC, 29208, United States, nhikmet@sc.edu 1 - Hierarchical Response Model For Casualty Processing In Mass Casualty Incidents Alkis Vazacopoulos, Optimization Direct, Inc., 202 Parkway, Harrington Park, NJ, 07640, United States, alkis@optimizationdirect.com, Nathaniel Hupert, Dimitris Paraskevopoulos, Panagiotis Petros Repoussis, Panagiotis Petros Repoussis This work presents a response and resource allocation model in the aftermath of a Mass-Casualty Incident. A mixed integer math programming formulation is proposed for the combined ambulance dispatching, patient-to-hospital assignment, and treatment ordering problem. The goal is to allocate effectively the limited resources during the response effort so as to improve patient outcomes, while the objectives are to minimize the overall response time and the total flow time required to treat all patients. The model is solved via exact and MIP-based heuristic methods. The applicability of the model and the performance of the new optimization methods are challenged on realistic scenarios. 2 - Key Factors And Patterns In Employee Choice Of High Deductible Health Plans Qing Ye, Purdue University, 315 N Grant St, West Lafayette, IN, 47907, United States, yqing@purdue.edu, Bhagyashree Katare, Yuehwern Yih U.S. employers have increasingly provided high-deductible health plans in response to the rising cost of health care. This study provides insight into selecting and switching behavior of employees towards high-deductible health plans. Data mining techniques are utilized to identify factors associated with their plan choice mobility. A case study is presented using five years’ claims information. 3 - Hospital Information Technology Investment Impacts On Patient Satisfaction, Clinical Performance, Efficiency, And Patient Outcomes Neset Hikmet, Associate Professor, University of South Carolina, This study extends prior research on healthcare information technology (HIT) investment impacts on hospital performance. We analyzed 102 different HIT investment types; categorized them as clinical, administrative, strategic, and infrastructure HIT investments; and examined relationships between these and hospital performance scores. Combining secondary survey data from U.S. hospitals and a separate data set from CMS, we found significant positive and negative relationships between clinical and infrastructure HIT investment and hospital total performance, clinical performance, patient satisfaction, and efficiency scores - controlling for organizational factors. WA76 Legends D- Omni Applied Probability I Contributed Session Chair: Julian Sun, Cornell, Ithaca, NY, 14850, United States, ys598@cornell.edu 1 - Flexible Estimation Of Conway-maxwell Poisson Distribution Suneel Babu Chatla, Doctoral Student, National Tsing Hua University, Hsinchu, 30013, Taiwan, suneel.chatla@iss.nthu.edu.tw, Galit Shmueli The Conway-Maxwell Poisson (CMP) distribution is popularly used for its ability to handle both overdispersed and underdispersed count data. Yet, there is no efficient algorithm for estimating CMP regression models, especially with high- dimensional data. Extant methods use either nonlinear optimization or MCMC methods. We propose a flexible estimation framework for CMP regression based on iterative reweighted least squares (IRLS). Because CMP belongs to the exponential family, convergence is guaranteed and is more efficient. We also extend this framework to allow estimation for additive models with smoothing splines. 1301 Gervais St, Suite 1010, Columbia, SC, 29208, United States, nhikmet@sc.edu, Benjamin Schooley

389