Informs Annual Meeting 2017

SC45

INFORMS Houston – 2017

3 - Considering Practical Ethics in Community Based Operations Research (CBOR) David T. Hunt, Oliver Wyman, One University Square, Princeton, NJ, 08540, United States, david.hunt@oliverwyman.com We like to think that CBOR involves the application of analytical methods to “do good” locally; and, most will agree these efforts are “doing good.” However, what is meant by “doing good” and how do we know when our models are not “doing good” and may in fact be causing harm? Some philosophers consider happiness as the only good, and utilitarianism is based on maximizing happiness. But maximizing happiness can create winners and losers, and are we really “doing good” as long as the pluses outweigh the minuses? This presentation will explore these topics using examples drawn from CBOR. 4 - A Multi-methodology Approach for Community-based Operations Research in a New World Heritage Site Valentina Ferretti, London School of Economics and Political Science, Houghton Street| London | WC2A 2AE, London, United Kingdom, V.Ferretti@lse.ac.uk This study develops a participatory multi-methodology intervention designed and deployed to support planning in a new World Heritage site, i.e. the vineyard landscape of Langhe, Roero and Monferrato in Northern Italy. The proposed framework uses Spatial SWOT Analysis and Multicriteria Decision Aiding (MCDA) in Phase 1 (problem identification - knowledge phase), Stakeholders’ Analysis and Spatial MCDA in Phase 2 (problem formulation - planning phase), and Choice Experiments during Phase 3 (problem solving - design). This talk focuses on the results and impacts associated with each phase of the process, by discussing the respective levels of community engagement and methodological innovation. 360C Healthcare Analytics Applications Sponsored: Public Sector OR Sponsored Session Chair: Andres Garcia-Arce, Geisinger Health Systems, Danville, PA, 17822, United States, garciampu@gmail.com 1 - Stacking Machine Learning Models for the Prediction of No-shows in Outpatient Neurological Appointments Andres Garcia-Arce, Geisinger Health Systems, 100 North Academy Avenue, MC 30-65, Danville, PA, 17822, United States, garciampu@gmail.com Patients not showing for their appointments hurt the access and quality of care for other patients. This study aims to improve predictive models for no-shows using machine learning algorithms, including ensembles of models. A novel aspect is that we consider cancelation within 7 days as a no-show. We use data from neurological outpatient appointments made in clinics from Geisinger Health Systems. This contribution will help improve the existing efforts in the reduction of no-shows, increase the access to care, and therefore, improve quality of care and patient satisfaction. 2 - Scenario-based Ambulance Deployment with Two Types of Servers Soovin Yoon, University of Wisconsin–Madison, 1415 Engineering Drive, Room 3261, Madison, WI, 53706, United States, yoon57@wisc.edu Ambulance deployment problem with multiple server types is challenging to solve due to inherent uncertainties in emergency calls as well as dependencies between servers. We propose a scenario-based ambulance deployment model with two types of servers, where the objective involves not only the response times but also the match between the server type and patient priority. The problem is formulated as a stochastic integer program with data-driven scenarios. Decomposition algorithms are applied to solve large size instances. 3 - Discovering Patterns in Self-reported Care Experience Data and Clinical Outcome Data to Improve Patient Experience Qianyu Hu, Penn State University, 126 E.Nittany Avenue, Apt 12, State College, PA, 16801, United States, qianyuhu312@gmail.com Over past 25 years, patient satisfaction has gained increasing attention. Research has shown that patient experience is critical to improving clinical outcomes and should be measured in addition to standard quality and safety metrics. Therefore, providing a better patient experience, measuring and reporting is regulatory. However, literature is divided on this. We study and analyze the relationship between patient satisfaction and clinical outcomes using statistical and graphical techniques. We use HCAHPS survey data to examine the relationship between patient perspective and clinical outcomes, to provide quality care. SC44

4 - A Predictive Model to Estimate Post Acute Care Needs: A Cohort Study of Patients with Coronary Artery Bypass Graft or Valve Replacement Ineen Sultana, Graduate Research Assistan, Texas A&M.University, College Station, TX, 77843, United States, ineensultana@tamu.edu Predicting Discharge Location (DL) for patients is essential for better care coordination with post-acute care (PAC) facilities; it not only aid in efficient discharge planning but also help the PAC facilities project the evolving demands. In this study, a retrospective cohort analysis of patients with coronary artery bypass graft or valve replacement is conducted, and a predictive model is developed using multinomial logistic regression to estimate the potential DL for a patient. Our analysis identified the hospital location, characteristics, and stay in hospitals as well as the characteristics and comorbidity of patients to be the most significant influential factors for the DL estimation. 5 - Real-Time Asset Distribution System: Deploying Machine-learning Solutions in the Emergency Department Hospitals are challenged to provide timely patient care while maintaining high resource utilization. Real-time asset management is one of such initiative whereby clinical staff convenes each morning to forecast low- to high-demand hospital units and ensure that the necessary equipment, such as IV pumps, are available for providing patient care. We designed and deployed a real-time asset distribution system that leverages on machine learning, electronic medical records, and real-time location system to guide clinical staff in recovering under- utilized equipment and relocate them where they are (or will be) needed. 360D JFIG Panel Discussion: Best Practices in Reviewing Papers Sponsored: Junior Faculty JFIG Sponsored Session Chair: Jose Luis Walteros, University at Buffalo, SUNY, 413 Bell Hall, Buffalo, NY, 14260, United States, josewalt@buffalo.edu 1 - JFIG Panel Discussion: Best Practices in Reviewing Papers Canan Gunes Corlu, Boston University, 808 Commonwealth Avenue, Boston, MA, 02215, United States, canan@bu.edu The panelists consist of senior professors with vast experience in reviewing papers for different journals, including Management Science, Operations Research, INFORMS Journal on Computing and Decision Analysis. The panel will share guidelines and best practices for paper reviewing. They will also answer questions pertaining to writing and publication. 2 - Panelist Jennifer Ryan, University of Nebraska-Lincoln, Lincoln, NE, United States, jennifer.ryan@unl.edu 3 - Panelist Jay Simon, American University, 4400 Massachusetts Avenue, NW, Washington, DC, 20016, United States, jaysimon@american.edu 4 - Panelist Laurens G.Debo, Dartmouth College, 100 Tuck Hall, Hanover, NH, 03755, United States, laurens.g.debo@tuck.dartmouth.edu 5 - Panelist James R.Wilson, North Carolina State University, Charlotte, NC, United States, jwilson@ncsu.edu Diego Martinez, Johns Hopkins University, Baltimore, MD, United States, dmart101@jhmi.edu, Jeffrey Appelbaum SC45

83