2016 INFORMS Annual Meeting Program

MA20

INFORMS Nashville – 2016

MA20 106C-MCC Healthcare Analytics: Big Data, Little Evidence Invited: Tutorial Invited Session Chair: Joris van de Klundert, Erasmus University, Institute of Health Policy & Management, 2983 HE Ridderkerk, Netherlands, vandeklundert@bmg.eur.n 1 - Healthcare Analytics: Big Data, Little Evidence Joris van de Klundert, Erasmus University, Institute of Health Policy & Management, 2983 HE Ridderkerk, Netherlands, vandeklundert@bmg.eur.n While the healthcare sector contributes more than ten percent of GDP in most developed countries and is approaching twenty percent in the US, it remains a relatively modest area in the field of Operations Research, Management Science, and Analytics. There is considerable room for a larger and more valuable contribution, especially in view of the important advancements in information technology taking place in healthcare across the globe, which are already contributing to reducing the global burden of disease. In order for Analytics professionals and scientists to reach the full contribution potential of their discipline, it is beneficial to understand the dominant research paradigms and results of clinical and health sciences research. These sciences are rooted in empirical evidence, in empirical data, thus offering connection opportunities. In this tutorial we review the current position of Analytics as covered in the Operations Research and Management Science literature, and outline a path for the science of Analytics to enlarge its contribution to the health of populations. Chair: Muge Capan, Christiana Care Health System, Value Institute, 4755 Ogletown-Stanton Road, 2nd Floor, Suite 2E55, Newark, DE, 19718, United States, Muge.Capan@ChristianaCare.org 1 - Sepsis: Sepsis Early Prediction Support Implementation System Nisha Nataraj, North Carolina State University, NC, United States, nnatara@ncsu.edu, Julie Ivy, Muge Capan, Ryan Arnold, James R Wilson Sepsis can be broadly defined as an infection plus systemic manifestations of the infection. It remains the most expensive condition in hospitals as well as one of the leading causes of in-hospital mortality. Using a discrete-event simulation framework, this research aims to develop personalized intervention policies for patients with sepsis. Specifically, we focus on the toll comorbidities and pre- existing conditions take on the manner in which sepsis presents and the associated impact they have on decision making. 2 - Validation And Implementation Of Early Warning System To Synthesize Acuity, Clinical Judgment, And Workload Stephen Hoover, Christiana Care Health System, Value Institute, Newark, DE, United States, Stephen.Hoover@ChristianaCare.org Muge Capan, Justin M Glasgow, Susan Mascioli, Eric V. Jackson We present the validation and implementation of a clinical early warning system. An analytical framework was developed to quantify physiological deterioration that results in adverse events by integrating a published Early Warning Score and a new Nurse Screening Assessment tool. Survival Analysis and Monte Carlo methods were used to validate our approach. Relevant system costs and workload implications were analyzed. Findings indicate potential for reduction in variation of care and prevention of unnecessary ICU transfers. Iterative implementation processes highlighted the importance of multidisciplinary teams and systemwide education when modifying complex healthcare systems. 3 - Patient Access To Specialty Endocrinology Care Henry Ballout, University of Michigan, Ann Arbor, MI, United States, haballou@umich.edu, Pranjal Singh, Amy Cohn, Amy E. Rothberg Recurrent patient visits add tremendous complexity to modeling capacity utilization for healthcare professionals. To assist an endocrinology clinic at the University of Michigan, we present a temporal database that takes prospective patient appointment and provider availability data and enables capacity analysis through a compilation of daily snapshots. Using this database, we investigate the clinics issues regarding access and adherence to the program’s highly structured timeline. MA21 107A-MCC Applications of Health Systems Analytics Sponsored: Health Applications Sponsored Session

4 - Robust Optimization Framework To Account For Prediction Errors For Cancer Diagnosis Selin Merdan, University of Michigan, smerdan@umich.edu Multiple diagnostic tests are often available for diagnosing diseases such as cancer; however, how best to use these tests to render a diagnosis is challenging because there is often a tradeoff between the benefits of diagnosis and the harms and costs associated with the diagnostic tests themselves. We present a robust optimization model for determining the optimal assignment of composite diagnostic tests based on individual patient risk factors to achieve an optimal balance between the benefits and harms of diagnostic tests. We further provide a specific example in the context of radiologic imaging to detect metastatic prostate cancer.

MA22 107B-MCC

Healthcare Policy, Personalized Treatment, and Coalitions: Operations Research Approaches Invited: ORinformed Healthcare Policies Invited Session Chair: Pooyan Kazemian, Harvard Medical School, 25 Shattuck Street, Boston, MA, 02115, United States, pooyan.kazemian@mgh.harvard.edu Co-Chair: Mark P. Van Oyen, University of Michigan, Ann Arbor, MI, United States, vanoyen@umich.edu 1 - Depression Care Management: Personalized Assessment To Cost-effective Population Interventions Shan Liu, Assistant Professor, University of Washington, Seattle, WA, 98195, United States, liushan@uw.edu Shuai Huang, Ying Lin, Xuelu Yang, Jiaqi Huang, Weiwei Shang Depression affects 1 out of 10 Americans. While electronic health record (EHR) provides an unprecedented information infrastructure, we need a system perspective and an associated computational platform, and a seamless integration with decision-analytic models to link the design of personalized disease interventions to cost-effective population management. We are developing methods to 1) analyze and predict the heterogeneous depression trajectories of a patient population from EHR, 2) characterize the hidden disease processes and design personal intervention schedules, and 3) evaluate their cost-effectiveness to David W. Hutton, University of Michigan, dwhutton@umich.edu In the last decade, high-cost biologics have taken center stage in the treatment of complex degenerative eye diseases. Biologics for ophthalmologic use consume one-sixth of Medicare’s part B drug budget. We examine how OR-based tools can be used to forecast and evaluate the national policy impact of these therapies (on both outcomes and costs). We see how different parties frame the problems differently and make different modeling choices. We discuss how different ways of using OR-based tools can influence policy. We explore open challenges evaluating the value of information and understanding how to evaluate decisions made jointly between patients and providers. 3 - On Personalized Allocation Of Treatments Mohsen Bayati, Stanford University, bayati@stanford.edu, Hamsa Sridhar Bastani, Khashayar Khosravi Growing availability of data has enabled practitioners to tailor medical treatments at the individual-level. This involves learning a model of decision outcomes conditional on individual-specific covariates or features. Recently, “contextual bandits” have been introduced as a framework to study these online decision making problems. In this talk we discuss statistical challenges that arise in such data-driven allocation of treatments. 4 - From Incident To Inpatient: How Healthcare Coalitions Can Improve Community Response monitor and treat depression across a range of care scenarios. 2 - National Policy Impact Of High-cost Biologics For Ophthalmologic Use Healthcare coalitions are a new type of organization that can coordinate casualty distribution among available hospitals in a metropolitan area after a multiple casualty incident. Using data from a major metro area, we show the value of different types of coordination regarding hospitals’ capacity information. We find that, while coalitions were initially create for large disasters, significant value comes in coordinate metropolitan area hospital resources on a much smaller scale. We also identify what types of information are most valuable for these organizations to collect and disseminate. This leads to interesting policy implication for the further funding and creation of such coalitions. Jonathan Helm, Indiana University, helmj@indiana.edu Alex Mills, Andres Jola-Sanchez, Mohan V Tatikonda, Bobby Courtney

128