Informs Annual Meeting 2017

TD04

INFORMS Houston – 2017

TD05

3 - Inferring the Causes Behind Fresh Food Spending Elisabeth Paulson, Massachusetts Institute of Technology, Cambridge, MA, United States, epaulson@mit.edu A growing number of strategies are being proposed and tested to increase consumption of healthy food in poor or underserved neighborhoods. While there is agreement that the food environment is linked to diet and health outcomes, there is surprisingly little agreement about the factors that cause this relationship, due in part to the observational and sparse nature of most data. Thus, arguments in favor of various strategies for increasing healthy food consumption are typically based on tenuous observed relationships. In this paper we analyze the impact of access to grocery stores and value of nutrition on fresh food spending among SNAP households using the Food Acquisition and Purchase Survey (FoodAPS) dataset. Using the methodology of causal inference, we are able to quantify and compare the causal effects of both access and value. We find that the effects of both are significant among certain populations, but that they impact household’s decisions differently. Finally, we discuss which strategies for increasing fresh food spending are promising among various populations. 320A MSOM Sustainable Operations Sponsored: Manufacturing & Service Oper Mgmt, Sustainable Operations Sponsored Session Chair: Isil Alev, Boston College, Chestnut Hill, MA, 02467, United States, alev@bc.edu 1 - Extended Producer Responsibility in Developing Economies – Design Incentives and Infrastructural Development TD04 Producers’ investment and effort in building recycling infrastructure can be observed in practice especially in developing countries where such infrastructure is missing. Motivated by these examples, this paper studies the cost efficiency and environmental impact such as producers’ eco-design incentives when infrastructure investment decisions are endogenized. 2 - An Analysis of Time-based Pricing in Electricity Supply Chains Asligul Serasu Duran, Kellogg School of Management, 825 W Cornelia Avenue, Apt GDN, Chicago, IL, 60657, United States, a-duran@kellogg.northwestern.edu, Baris Ata, Ozge Islegen This study builds a framework for the retail electricity market to empirically evaluate the impact of time-based tariffs on the electricity supply chain. We find that optimal time-based tariffs reduce peak demand, but do not change consumers’ total demand or electricity bills significantly. Time-of-use tariffs with predetermined rates can capture most of the benefits of real-time prices. The environmental impact of time-based tariffs depends on the characteristics of the electricity market under study. 3 - Fair Labor Practices and Increasing Returns to Scale Susan A. Slotnick, Cleveland State University, OSM Department, BU 542, Cleveland, OH, 44115, United States, s.slotnick@csuohio.edu, Matthew J.Sobel This paper investigates the tradeoffs involved in a firm’s efforts to ensure that its products are manufactured using fair labor practices. We use a Markov decision process model of a firm that sources from one supplier and decides whether to cut its profit margins and/or expend resources to ensure fair labor practices at the supplier. Retail demand is influenced by reputation: if unfair labor practices come to light, reputation deteriorates, hence demand and profit diminish. Analytical and computational methods yield insights into the firm’s profit maximizing behavior, including the result that the firm experiences increasing returns to scale as a function of its reputation. 4 - On Recycling of Post-Disaster Debris Andriy Shapoval, Georgia Tech, 1449 Willow Lake Dr NE, Apt B, Atlanta, GA, 30329, United States, ashapoval3gt@gmail.com, Pinar Keskinocak, Beril Toktay This work studies emerging recycling policies regarding post-disaster debris management. We develop a game-theoretic model of sequential decision makers utilizing debris removal funding, including state and local government and recyclers, and analyze the total welfare of policies incentivizing recycling. We determine conditions for effective policies under conflicting stakeholder preferences. Luyi Gui, The Paul Merage School of Business, UC Irvine, University of California, Irvine, Irvine, CA, 92697-3125, United States, luyig@uci.edu

320B Data Analytics in Healthcare Sponsored: Health Applications Sponsored Session Chair: Mahboubeh Madadi, Louisiana Tech University, Ruston, LA, 71272, United States, madadi@latech.edu 1 - Bi-level Heterogeneous Degradation Modeling of Functional Performance for the Aging Population Quantifying heterogeneity among the elderly with functional performance degradation is crucial for effective long-term care planning and delivery. Existing studies mainly model degradation heterogeneity at either sub-population or individual level. This work proposes a bi-level heterogeneity modeling and quantification framework. At sub-population level, a Bayesian non-parametric model is proposed to relax the assumption of pre-specifying the number of sub- populations and realize the joint model estimation and selection. At individual level, functional data analysis tools are considered to extract temporal degradation signatures with risk/protective factors identified. 2 - Identifying Smoking Status from User-generated Content in an Online Community for Smoking Cessation Xi Wang, University of Iowa, 810 W Benton St B213, Iowa City, IA, 52246, United States, xi-wang-1@uiowa.edu, Kang Zhao Online communities for smoking cessation are increasingly popular sources of information and support for smokers who want to quit. The ability to automatically find smoking status from user-generated content in such an online community can help to provide better interventions. This study developed a machine-learning-based approach to identify individuals’ abstinence from their posts. Augmenting the traditional bag-of-words model, our approach leveraged three new types of features: context-specific features, author-based features, and thread-based features. The evaluation outcomes show that the three types of new features all contribute to the better classification performance. 3 - Detecting Patient-specific Abnormal Synchronization Pattern of Epileptic Seizures Miaolin Fan, Northeastern University, Boston, MA, United States, fan.mi@husky.neu.edu, Chun-An Chou In this study, we propose an efficient approach to detecting abnormal synchronization phenomena of epileptic seizures in a multivariate EEG system. The proposed method aims at characterizing the nonlinear dynamics of spatio- temporal EEG recordings using recurrence network and spectral graph theory. We present the computational results for CHB-MIT Scale EEG Epilepsy Database in various experimental settings, which validate the efficiency of detecting seizure onsets, as well as robustness to the noisy EEG data. 4 - Inpatient Discharge-by-Noon: Are Fewer Better Than All? Pratik Parikh, PhD, Wright State University, Dayton, OH, United States, pratik.parikh@wright.edu, Nicholas Ballester Discharge before noon has rapidly emerged as an unspoken golden standard in hospitals nationwide. While research has indicated a strong link between earlier inpatient discharges and better patient outcomes, especially upstream patient boarding, quantitative analysis of the expected benefits of discharging all patients before noon is lacking. Building upon our previous work, we explore the expected benefits of discharging fewer or all medically-ready patients by noon. Our findings suggest the law of diminishing returns applies; benefits decrease with an increase in the number of such patients. Nazmus Sakib, INFORMS.chapter at University of South Florida, 2319 Campus Lake CT, Apt 204, Tampa, FL, 33612, United States, nsakib@mail.usf.edu, Xuxue Sun, Nan Kong, Hongdao Meng, Mingyang Li

376