INFORMS 2021 Program Book

INFORMS Anaheim 2021

ME22

3 - Optimal Presentation of Alternatives Zeya Wang, Georgia Institute of Technology, Atlanta, GA, United States, Morvarid Rahmani, Karthik Ramachandran In many contexts such as healthcare, knowledge outsourcing, or product design and development, a provider may have multiple alternatives that could potentially solve the client’s problem. A key decision for the provider is: how to present these alternatives to a client? In this paper, we develop a game-theoretic model where the provider chooses which alternative to present and in what sequence, and the client chooses which alternative to implement. We characterize the optimal strategies for the provider in equilibrium to determine which alternative the provider should offer and when to offer it. We study the effects of implementation ability, asymmetric implementation cost and correlation between options on these strategies. In Person: Healthcare Analytics in Action: Using Data-driven Models to Effect Change General Session Chair: Retsef Levi, MIT, Cambridge, MA, 02142-1320, United States Co-Chair: Taghi Khaniyev, MIT Sloan School of Management, Cambridge, MA, 02142-1508, United States 1 - Optimal Resource Pooling for Effective Use of Future Operating Room Capacity Seung-Yup Lee, Vanderbilt University Medical Center, Nashville, TN, 37212, United States, Vikram Tiwari We investigate the optimal timing for release of unfilled operating room (OR) block capacity to improve the efficiency and effectiveness of OR use. A Markov decision process structure is designed that incorporates not only the number of remaining days until the day of surgery but also the remaining capacity of the OR block as well as historical demand for the block. In this presentation, we propose the decision-making modeling structures for both the single- and multi-OR block cases and discuss the applicability of the resulting polices in practice and expected improvements. Our results provide insights into pursuing proactive management of pooling limited resources in the healthcare setting where both system-wide efficiency and specialty-specific characteristics of resources should be considered. 2 - A Prescriptive Approach to Surgical Inpatient Discharges Taghi Khaniyev, MI T. Sloan School of Management, Cambridge, MA, 02142-1508, United States, Kyan Safavi, Martin Copenhaver, A. Cecilia Zenteno, Bethany Daily, Peter Dunn, Retsef Levi We first trained a neural network model to accurately predict next-day’s inpatient discharges using structured EHR data which was represented based on whether it indicated a clinical or administrative barrier to discharge which was defined as an event that may postpone the patient’s discharge. Discharge predictions were categorized as NO/MAYBE/YES. An optimization model was developed to select the minimal subset of barriers for each patient that need to be resolved in order to move a patient to YES category. This minimal list was intended to serve as a prioritized action list for each patient. When we augmented the prediction model with free-text clinical notes using a recurrent neural network, the prediction accuracy was improved by up to 20%. ME24 CC Room 205A In Person: Emerging Topics in Sustainable Operations General Session Chair: Adem Orsdemir, University of California-Riverside, Riverside, CA, 92521-9800, United States 1 - Is Adopting Mass Customization a Path to Environmentally Sustainable Fashion? Adem Orsdemir, University of California-Riverside, Anderson Riverside, CA, 92521-9800, United States, Aydin Alptekinoglu In high-product-variety businesses like fashion, mass production systems create environmental waste in the form of overproduction on a colossal scale. Mass customization has been proposed — without solid evidence — as a solution. In this paper, we analyze whether mass customization can indeed offer a win-win solution that helps both the bottom line and the environment. We also study the impact of three real policy options: promoting mass customization, charging a disposal fee for overproduction, and recycling. ME22 CC Room 204B

2 - How Does Physical Access Affect Emergency Department Utilization? Evidence From Insurance Coverage Expansion Eric Xu, University of Minnesota, Minneapolis, MN, 55455-0438, United States, Anant Mishra, Kevin Linderman The Patient Protection and Affordable Care Act was an attempt to provide widespread insurance coverage. While the law’s Medicaid Expansion provided individuals with a financial means, we find that the impact of physical accessibility, e.g. spatiotemporal characteristics, has a non-trivial impact on emergency department use. We also find that congestion at the nearest primary care clinic due to a post-enforcement increase in Medicaid managed care encounters represents an underlying mechanism affecting annual emergency department use. 3 - Empirical Investigation of Locational Demographics and Facility Emissions Abhinav Shubham, PhD Student, Georgia Institute of Technology, Atlanta, GA, United States, Ravi Subramanian Environmental Justice encapsulates the idea of fairness in protection for communities from environmental and health hazards, regardless of race, color, national origin, or income. It is relevant to the practice of OM in the form of inequities that result from disparate operational decisions or policies. The research question that we aim to address is: How disparate are facility-level emissions across communities with different racial makeups? To address this question, we draw data multiple sources. we address the confounding effects of pre-treatment factors through various methods to assess how facilities may differ in their emissions (outcome) between locations that differ in racial makeup. Our findings offer evidence for regulatory intervention and opportunities for firms to reconsider their ESG objectives with local considerations of fairness and equity. ME25 CC Room 205B In Person: Julia Packages for the Modeling and Solution of Optimization Problems General Session Chair: Joshua Pulsipher, University of Wisconsin Madison, United States 1 - Strengths of Approximations for Bipartite Bilinear Programs Akshay Gupte, University of Edinburgh, Edinburgh, United Kingdom We present theoretical analyses of some inner and outer approximations of a bipartite bilinear program. The outer approximations we consider are from standard methods such as Reformulation Linearisation Technique and Lagrange Duality. In particular, we give a combinatorial upper bound on the relative gap of the RLT to the optimum by showing that this gap is bounded by the chromatic number of a certain graph that is obtained from the co-occurrence graph of the problem. Secondly, we give a sufficient condition for the RLT or a lifting of it to be exact (i.e., relative gap of 1). Thirdly, we show that convexifying all the bilinear terms simultaneously over the domains of the variables is equivalent to taking a certain semi-Lagrangian of the problem. The combinatorial bound from the first result on RLT also allows us to bound the gap from a MIP inner approximation of the problem. 2 - InfiniteOpt.jl: A Unifying Abstraction for Infinite-Dimensional Optimization Joshua Pulsipher, University of Wisconsin-Madison, Madison, WI, United States, Weiqi Zhang, Victor M. Zavala Infinite-dimensional optimization problems are a challenging problem class that cover a wide breadth of optimization areas and embed complex modeling elements such as infinite-dimensional variables, measures, and derivatives. Typical modeling approaches (e.g., those behind Gekko and Pyomo.dae) often only consider discretized formulations and do not provide a unified paradigm across the various disciplines. We present InfiniteOpt.jl which facilitates a coherent unifying abstraction for characterizing these problems rigorously through a common lens. This decouples models from discretized forms and promotes the use of novel transformations. This new perspective encourages new theoretical crossover and novel problem formulations (creating new disciplines like random field optimization).

94