Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MA57

4 - A Structured Solution of Prescriptive Analytics: Theory and Experiment Justin Jia, Purdue University, West Lafiette, IN, United States, Elena Katok Prescriptive analytics problem is when a decision maker solves an optimization problem with uncertainty without knowing the distribution of the random variables, but has a sample. Our insight is that we show that the decision, which is a high-dimensional function of the sample, can be reduced to a simple single- dimensional function of a sufficient analytic statistic. We report on a controlled laboratory experiment with human subjects, and show that behavior qualitatively matches theoretical predictions. n MA55 North Bldg 232C He for She: What Does it Mean for Us at INFORMS Sponsored: Women in OR/MS (WORMS) Sponsored Session Chair: Dorothee Honhon, University of Texas at Dallas, Richardson, TX, 75080, United States Co-Chair: Margarit Khachatryan, MagAnalytics, Bridgeton, MO, 63044, United States 1 - Moderator Margarit Khachatryan, Founder, MagAnalytics, St. Louis, MO, 63044, United States In this panel session, a group of influential male speakers will discuss how we can best support women in OR/MS. They will share their experiences working with/for female colleagues, and supervising/mentoring them. They will discuss family-focused, anti-discrimination and sexual harassment institutional policies and share tips on how to make our work places a friendly and safe environment for women and other minorities. Panelists Rohit Verma, NY, United States Anton Ovchinnikov, Queen’s School of Business, 143 Union Street, Kingston, ON, K7L 3N6, Canada Richard G. McGrath, United States Naval Academy, Annapolis, MD, United States Rishabh Bhandawat, University at Buffalo, 25HA Creekside Village, Buffalo, NY, 14261, United States Leon McGinnis, Georgia Institute of Technology, Isye Dept, 755 Ferst Drive, Atlanta, GA, 30332-0205, United States Gregory James King, GAP, San Francisco, CA, United States n MA56 West Bldg 101A Joint Session DM/HAS: Patient-Centered Health Care Sponsored: Health Applications Sponsored Session Chair: Guihua Wang, Ann Arbor, MI, 48105, United States 1 - Driving Personalized Health Care through Heterogeneous Data Analysis Guihua Wang, Ross School of Business, University of Michigan, 701 Tappan Street, Ann Arbor, MI, 48105, United States, Jun Li, Wallace J. Hopp This study addresses the challenges of generating patient-centric outcome information. Using patient-level data from thirty-five hospitals for six cardiovascular surgeries in New York State, we identify patient groups that exhibit significant differences in outcomes with a new instrumental variable tree approach. We find that outcome differences between hospitals are heterogeneous not only across procedure types, but also along other dimensions such as patient age and comorbidities. We illustrate how patient-centric information can help patients make more informed decisions, payers enhance pay-for-performance programs, and providers target quality improvement efforts. 2 - An Analytics Approach to Predicting Treatment Effectiveness: Inter-hospital Transfer of Heart Attack Patients Susan F. Lu, Purdue University, Krannert 441, West Lafayette, IN, 47907, United States, Qi Feng, George Shanthikumar Healthcare regionalization is an important topic for the on-going healthcare reform. In this study, we integrate healthcare domain knowledge based structural model into data analytics and apply the method to study the inter-hospital

transfer of heart attack patients. We developed a two-stage support vector machine model to identify heart attack patients who need to be transferred immediately given the limited revascularization resources. Our recommendation performed far better than physicians’ decisions in terms of expected lives saved. 3 - Managing the Portfolio of Elective Surgical Procedures Hessam Bavafa, Wisconsin School of Business, 4284C Grainger Hall, 975 University Avenue, Madison, WI, 53706, United States, Sergei Savin, Lerzan E. Ormeci We consider the problem of allocating daily hospital service capacity among several types of elective surgical procedures. Our focus is on the interaction between two major, constraining hospital resources: operating room and recovery bed capacity. In our model, each type of surgical procedure has an associated revenue, stochastic procedure duration, and stochastic length-of-stay (LOS). We consider arbitrary distributions of procedure and LOS durations and derive a two- moment approximation for the total procedure duration and the daily number of occupied beds for a given portfolio of procedures. 4 - Using Episodes of Care to Examine Mental Healthcare Quality Jayakanth Srinivasan, Questrom School of Business, Boston, MA, United States Episodes of care are useful for integrating the discrete, itemized world view obtained from traditional healthcare utilization data. In this paper, we show how care trajectories can be constructed for Soldiers using de-identified data, and calculate gross measures of resource utilization in the outpatient and inpatient settings at two Army hospitals. I compare two approaches for constructing episodes of mental healthcare within these care trajectories, the first using only administrative data, and the second incorporating heuristics obtained from field research on health systems operations. The approaches developed in this paper provide a new lens for examining quality and patient safety. Emergency Management Sponsored: Health Applications Sponsored Session Chair: Douglas R. Bish, Virginia Tech, Blacksburg, VA, 24061, United States Co-Chair: Paul Bartholomew, VA, United States 1 - Multi-class Casualties Distribution in Mass Casualty Incidents Zheqi Zhang, University of North Carolina, Chapel Hill, NC, United States We study the patient distribution problem in the mass casualty incidents aims to maximize the expected number of survivors. Our Markov decision process(MDP) model assumes patients triaged into two classes distinguished by survival probabilities and medical needs. The decision is to prioritize and distribute patients by limited ambulances to hospitals with different capabilities and capacities. We explicitly modeled the ED and operating room capacities to capture the possible congestion. Base on this MDP formulation, we propose heuristic policies and employ discrete-event simulations to demonstrate the benefits of using the proposed heuristics against benchmark policies from practice. 2 - Coordinating Regional Response to a Mass-casualty Incident Paul Bartholomew, Virginia Tech, Blacksburg, VA, 24060, United States, Douglas R. Bish, Behrooz Kamali We study optimally coordinating the response to a mass-casualty incident. An effective response requires coordination of several entities including first responders and hospitals in the region. We develop a model that determines the service order for the casualties and hospital assignment, given hospital and transportation resources. The objective of the model is to maximize the total survival probability using a novel function that accounts for type of the casualty, travel distance, and receiving hospital’s quality of care. We compare the results from our model to common real-world policies and relevant studies in the literature and analyze the performance under various resource settings. 3 - Emergency Preparedness and Response for Mass Gatherings Mohammadreza Torkjazi, Graduate Research Assistant, West Virginia University, Morgantown, WV, United States, Behrooz Kamali A mass gathering (MG) is an event where a large crowd gathers. This creates a risk of delayed and/or limited response to emergencies. In this research, we propose a framework to quantify and measure this risk by identifying factors affecting the risk and classify them into multiple categories. Then, using “controllable” factors in each category, we develop a mathematical model to minimize the risk exposure to casualties by improving the flow of response. We present several case studies to demonstrate the effectiveness of our approach and analyze its sensitivity to various factors. n MA57 West Bldg 101B

139