Informs Annual Meeting 2017

TC06

INFORMS Houston – 2017

TC06

2 - Evaluating the Performance of Emergency Department Physicians Raha Imanirad, Harvard University, Boston, MA, United States, rimanirad@hbs.edu Given the vital role of emergency department (ED) physicians in the timely and effective delivery of care, it appears pertinent to have a measure of physician performance in place. Despite recent advances, however, a standard metric for evaluating ED physicians has yet to emerge. In this study, we use data envelopment analysis (DEA) to evaluate the performance of ED physicians based on two measures of quality and speed. In addition to identifying which physicians demonstrate best practices, we conduct various statistical analyses to examine which variables are associated with higher levels of physician performance. 3 - Measuring Emergency Care Networks with Location Allocation Models and GIS Peter Vanberkel, Dahousie University, P.O. Box 1000, Halifax, NS, B3 H.4R2, Canada, peter.vanberkel@dal.ca The objective of this study is to measure how the Nova Scotian Emergency Medical System (EMS), consisting of Emergency Departments (EDs) and ambulance services, covers the population. This study uses location-allocation models including the P-median problem (PMP) and P-centre problem (PCP). We analyze the existing network by computing the weighted travel distance of the population to each facility, the maximum distance any person must travel, and the number of people within a specified distance from a facility. We also consider several proposed changes to the network and compute the degree of improvement expected using these metrics. 322B Disaster and Emergency Management Contributed Session Chair: Zeina Bousaid, Lamar University, Beaumont, TX, United States, zbousaid@lamar.edu 2 - An Emergency Logistics Distribution Approach for Real-time Response to Relief Demand in Uncertain Environment Yajie Liu, Associate Professor, National University of Defense Technology, NUDT, Changsha, 410073, China, liuyajie@nudt.edu.cn, Zhiyong Wu, Yue Wang, Jianmai Shi When making emergency distribution plans in response to disasters, we should not only consider uncertain and dynamic input data but also the real-time adjustment requirements of the existing plans. A multi-commodity, multi-period distribution model is presented to minimize the total weighted unmet demands. Moreover, a rolling horizon-based framework that is based on the robust optimization and model predictive control approaches is proposed to obtain robust relief distribution plans and adjust them in accordance with updated real- time information. A numerical example based on the Great Sichuan Earthquake is utilized to investigate the application of our proposed model and approaches. 3 - Integrated Logistics Decisions and Representative Objectives of Preparedness and Response Phases of Disaster Operations Syed Tariq, Lahore University of Management Sciences (LUMS), Shah House, A-10, Block 2, Chapal Sun City, Lahore, 75280, Pakistan, 14080009@lums.edu.pk, Muhammad Naiman Jalil Disaster response operations are linked to preparedness. The response stage consists of three sub-stages: 1) Emergency response, 2) Continuum response, and 3) Initial recovery response (Sheu and Pan, 2016). In this paper, we present a model which utilizes objective functions which better represent the objectives of Humanitarian actors at each stage. The stages are modeled in conjunction with each other to see the impact of decisions made at each stage on overall performance. 4 - Determining Inventories of Relief Supplies During Disaster Response Operations by Predicting Non Linear Human Behavior with a System Dynamics Model Zeina Bousaid, Lamar University, 4400 SMLK Jr. Pkwy, P.O. Box 10643, Beaumont, TX, 77710, United States, zbousaid@lamar.edu, Alberto Marquez Forecasting relief supplies is a challenge due to uncertainties, limited resources, delays, disruptions, and complex human factors. We propose system dynamics to model people’s behavior during evacuation, and to forecast the relief inventory needed during disaster response operations. Our model consists of four compartments communicating through feedback loops and generating complex nonlinearities. FEMA Hurricane Evacuation reports are used for numerical validation, along with regression analysis, and Monte Carlo simulation. Risk is quantified, and policies are suggested for better response. TC08

320C Improving Access and Outcomes in Healthcare Sponsored: Health Applications Sponsored Session Chair: Maria Esther Mayorga, North Carolina State University, Raleigh, NC, 27695, United States, memayorg@ncsu.edu 1 - Resource Allocation Strategies under Dynamically Changing Patient Health Conditions Siddhartha Nambiar, North Carolina State University, 1120 Clarion Heights Lane, Apt 303, Raleigh, NC, 27606, United States, snambia@ncsu.edu, Maria Esther Mayorga Consider a group of patients being treated for a disease (such as Sepsis), whose condition may change over time. Resources allocated to a patient (e.g. bedside lactate) influence their disease progression and outcomes. In this study, our goal is to allocate limited number of resources to patients, depending on their health states, in order to maximize outcomes. We formulate two models; an MDP model with resources allocated dynamically, and a jackson queueing network, where the number of resources is assigned in advance. The performance of these policies is Gabriel Zayas-Caban, University of Wisconsin, Madison, WI, United States, zayascaban@wisc.edu, Amy Cochran, Keith Kocher The Hospital Readmissions Reduction Program penalizes hospitals with excess re- admissions. Re-evaluating admissions decisions, providers have considered delays or increases in decisions. To understand the impact of these changes on patient outcomes, we analyzed over 150,000 encounters in the Emergency Department from the University of Michigan from 2012 through 2015. A Hidden Markov model is fit to the data to capture the interplay between confounding latent variables (e.g. patient severity) and admission decision process. This allowed us to gain insight into how decisions are made and how changes in decisions impact patient outcomes. 3 - Strategies for Improving Specialty Care Appointment Scheduling Ashley N. Anhalt, University of Pittsburgh, 525 S. Aiken Avenue, APT.3, Pittsburgh, PA, 15232, United States, ana88@pitt.edu, Arnab Bhattacharya, Jeffrey P. Kharoufeh In resource-constrained environments, patients are often unable to schedule specialty care appointments within a reasonable time. We first present a stochastic optimization model that aims to maximize the proportion of patients who are scheduled within a prescribed time threshold, subject to provider availability. However, due to intractability of the model, we subsequently devise and analyze a set of computationally-efficient strategies that significantly reduce appointment scheduling delays. Our strategies are illustrated for a major healthcare provider’s cardiology unit. 4 - Microsimulation Model using Christiana Care Early Warning System (CEWS) to Evaluate Physiological Deterioration Shengfan Zhang, University of Arkansas, Fayetteville, AR, United States, na, Bin Li, Muge Capan Patients often exhibit signs of physiologic deterioration in the hours before harm events occur. Christiana and Wilmington Hospitals are adapting the Christiana Care Early Warning System (CEWS) to detect all-cause acute clinical deterioration impacting inpatients and support clinical decision making. We used a microsimulation model to simulate the inpatients cohort’s CEWS value and to explore the possible intervention policies on mortality, length of stay, average percent time over critical CEWS thresholds with the change of policies. 322A Operations Research for Critical Care Sponsored: Health Applications Sponsored Session Chair: Vedat Verter, McGill University, Montreal, QC, H3A 1G5, Canada, vedat.verter@mcgill.ca 1 - Dynamic Patient Location in the ICU Hamidreza Eslami, McGill University, Montreal, QC, Canada, hamidreza.eslami@mail.mcgill.ca Healthcare location studies at patient level usually target the groups of patients and not the individual ones. Due to the critical status of the ICU patients, their location on the ICU may have significant impacts on the unit operations and costs. We develop a model to optimally locate the incoming ICU patients on the floor so that the overall nurse costs and patient relocation costs are minimized by considering the evolvement of patients’ health status. Therefore, the patients who are prone to get better at the same time and thus share a nurse, will be located in adjacent rooms. then tested against existing practices and heuristics. 2 - Evaluating Hospital Readmission Decisions TC07

338