INFORMS 2021 Program Book

INFORMS Anaheim 2021

TC21

3 - A Dynamic Optimization Framework for Integrated Fossil Energy, Renewables and Flexible Carbon Capture for Transitioning Towards Clean Energy Manali S. Zantye, Texas A&M University, College Station, TX, United States, Akhil Arora, M.M. Faruque Hasan We address the intermittency of renewables and the high cost of CO2 capture through integration with existing fossil power plants and exploration of operational synergies for grid decarbonization. A mathematical optimization- based framework and a two-stage solution strategy is developed to evaluate if the benefits obtained from integration outweigh the investment cost under spatiotemporal variability of electricity markets and renewables. The results show that it is economically beneficial to integrate solar-assisted CO2 capture with one- third of the coal plants in the US to reduce emissions. TC19 CC Room 203A In Person: Data-Driven Decision-Making in Healthcare Applications General Session Chair: Maryam Alimohammadi, University of Arkansas, Fayetteville, AR, 72701, United States 1 - WiFi Network Logs Enable Data-driven Closure Policies for COVID-19 on University Campuses Lauren N. Steimle, Georgia Institute of Technology, Atlanta, GA, 30308, United States, Jingyu Li, Meghan Meredith, Dima Nazzal Universities commonly control infectious diseases like COVID-19 by employing broad closures, which come at the cost of learning outcomes and mental wellbeing concerns. To meet infection reduction goals while minimizing burdens, universities need a flexible approach to design and assess closure policies. We demonstrate that universities can conceive such localized closure by modeling campus mobility with anonymized WiFi logs already collected in their infrastructure. Using an agent-based model, we show that policies designed using these data can outperform simple remote instruction policies in terms of mobility reduction and disease outcomes. 2 - Prediction of Mechanical Ventilation Outcome in Intensive Care Units Using Modified Recurrent Neural Network Maryam Alimohammadi, University of Arkansas, Fayetteville, AR, 72701, United States, Shengfan Zhang Mechanical ventilation is one of the most widely used interventions in ICU. Because of the high usage and limited sources, the management of mechanical ventilation is significant. Predicting the ventilation outcome before the start of ventilation or at the beginning of the ventilation can improve decision-making in mechanical ventilation interventions in the ICU. However, dealing with the electronic health record data of patients admitted to ICU is complicated due to its complex temporal nature, noisiness, and irregularity. We proposed a modified recurrent neural network to predict the mechanical ventilation outcome using optimized decision-making time windows. This method helps reduce the missingness and noisiness of data and predicts the outcome with higher accuracy in comparison to the traditional classification methods. 3 - Evidence of Worse Outcomes Related to Out-of-Hospital Cardiac Arrest During the COVID-19 Pandemic Due to Patient Reluctance to Seek Care Christopher Sun, Massachusetts Institute of Technology, Boston, MA, 02114-3756, United States, Sophia Dyer, James Salvia, Laura Segal, Retsef Levi Delays in seeking emergency care stemming from patient reluctance may explain the rise in cases of out-of-hospital cardiac arrest (OHCA) and associated poor health outcomes during the coronavirus disease 2019 (COVID-19) pandemic. In this talk, we will discuss how we used emergency medical services (EMS) call data from the Boston area to describe the association between patient reluctance to call EMS for cardiac-related care and both excess OHCA incidence and OHCA- related outcomes during the COVID-19 pandemic.

TC20 CC Room 203B In Person: OR Applications for Medical Decision-making General Session Chair: Daniel Felipe Otero-Leon, University of Michigan, Ann Arbor, MI, 48109, United States 1 - Optimizing Repeated Decisions in Infectious Disease Control Suyanpeng Zhang, University of Southern California, Los Angeles, CA, United States, Sze-chuan Suen The Covid-19 outbreak emphasizes the necessity of studying policies to prevent or control the transmission of infectious diseases. However, evaluating dynamic policies for disease modeling can be challenging due to the complexity of the state space and transitions associated with realistic models of transmissible disease. We develop a Markov decision process framework for optimizing repeated decisions for infectious disease control over a population over time while considering uncertainty in parameters. 2 - Delivering Preventive Dental Care in Florida Schools: understanding System Limits Amin Dehghanian, Georgia Institute of Technology, Atlanta, GA, 30329, United States, Simin Ma, Yasin Cagatay Gultekin, Nicoleta Serban, Scott Tomar We evaluate the dental care availability to deliver preventive care for children in Florida elementary schools, with inferences on the availability limits to meet demand across all school. For this purpose, we use a bi-level optimization model to reallocate dentists’ caseload to school-based programs to meet the need for preventive dental care under a series of access and supply constraints. 3 - Patient Prioritization Model with Limited Resources and Stochastic Compliance Daniel Felipe Otero-Leon, PhD Student, University of Michigan, Physicians seek to prevent chronic diseases by tracking the patients’ healthcare behavior. They handle a large heterogeneous panel of patients, which is costly or demands multiple resources. Unfortunately, physicians do not count on infinite resources, for which they need to prioritize the panel. Not prioritize patients may forgo needed treatment and suffer adverse events related to the disease. Additionally, despite the physician’s efforts, prioritize patients may not adhere to medication and follow-up recommendations. We present a Multi-armed bandit model to maximize the panel’s total life years gain. Further, we tested the model using longitudinal data for cardiovascular diseases in a large cohort of patients seen in the national Veterans Affairs health system. Finally, we study the resulting prioritization policies and their structure. TC21 CC Room 204A In Person: Nonlinear Optimization Techniques in Stochastic Optimization General Session Chair: Ashish Chandra, Krannert School of Management, Purdue University 1 - Robust Optimization Approaches to Incentivize Carbon Footprint Reduction Aurelie Thiele, Southern Methodist University, Dallas, TX, 75275- 0123, United States It has become very important for many sustainability-conscious individuals and companies alike to reduce their carbon footprint (CO2 emissions), from operations to shipping to travel. We investigate tractable choice models with uncertain coefficients for incentivizing operational modalities for a heterogeneous population, provide theoretical insights into the tradeoffs between modalities and apply our approach to a real data set. 1205 Beal Avenue, Ann Arbor, MI, 48109, United States, Mariel Sofia Lavieri, Brian T. Denton, Jeremy Sussman, Rodney Hayward

117