Informs Annual Meeting 2017

MD07

INFORMS Houston – 2017

3 - Diagnosing Variations of Length of Stay through Integrating Control Charts and Lean Management Practices within the Crisp DM Framework Fadel Mounir Megahed, Miami University, Oxford, OH, fmegahed@miamioh.edu, Nasibeh Azadeh-Fard, Fatma Pakdil In-patient length of stay (LOS) at hospitals is considered a critical metric because it is widely used to measure the overall efficiency of healthcare services. Given the fact that abnormal LOS results in increasing cost both for hospitals and patients, this study proposes a new methodology, based on CRISP-DM framework including control charts and lean management understanding, for monitoring and improving LOS. The utility of the proposed method is demonstrated through a case study involving 11,722 internal medicine patients at a large hospital in Southwest Virginia. 4 - Simulation of Interdependent Operations in Outpatient Clinics: A Case Study Vahab Vahdat, PhD Candidate, Northeastern University, 334 Snell Engineering, 360 Huntington Avenue, Boston, MA, 02215, United States, vahdatzad.v@husky.neu.edu, Jacqueline Griffin Many patients require multiple services provided by different clinics during one visit to a health center. In this research an example of interdependencies among three outpatient clinics in Massachusetts, USA are explored. While orthopedic and rheumatology outpatient clinics both work independently, their operations are interdependent with the radiology clinic that provides imaging services for both clinics. In order to improve patient length of stay, a discrete-event simulation model is constructed to model day-to-day operations of each clinics and evaluate the impact of integrated decisions in patient scheduling and resource allocation. 322A Models and Algorithms for Scheduling Healthcare Providers Sponsored: Health Applications Sponsored Session Chair: Amy Cohn, University of Michigan, Ann Arbor, MI, 48109, United States, amycohn@umich.edu 1 - Algorithms to Generate Annual Rotation Schedules for Medical Residents William Pozehl, University of Michigan, Ann Arbor, MI, United States, pozewil@umich.edu, Amy Cohn An ever-growing list of educational requirements and service coverage needs make generating high-quality annual rotation schedules for medical residents increasingly difficult. On top of these rules, program leadership wishes to satisfy as many of the residents’ requests as possible to reduce the risk of burnout and prepare the residents for their desired career paths. We explore algorithms to generate schedules that not only meet the needs of the residents and services under these complex conditions, but also do so rapidly. 2 - Optimal Bed Assignment in Multi-occupancy Facilities Ruben Proano, Associate Professor, Rochester Institute of Technology, Gleason College of Engineering, 81 Lomb Memorial Drive, Rochester, NY, 14623, United States, rpmeie@rit.edu, Brenden M. Hoff In many parts of the globe, hospitalizations take place in multi-occupancy rooms. The need to admit as many patients as possible, and mitigate the risk of exposing inpatients to infections carried by other patients results in a constant shuffle of patients across different rooms. This study proposes a multi-stage optimization model that minimizes the impact of patient shuffles, and maximizes unit occupation. Experimentally, the model is also used to illustrate the effect of relying on batch bed-assignments to mitigate the need of shuffles. 322B Stochastic Models in Public Policy Sponsored: Health Applications Sponsored Session Chair: Ebru Korular Bish, Virginia Tech, Blacksburg, VA, 24060, United States, ebru@vt.edu 1 - Using Genetic Information to Guide Cholesterol Treatment Planning Wesley Marrero, University of Michigan, Ann Arbor, MI, United States, wmarrero@umich.edu, Suzanne Butler, Jeremy Sussman, Mariel Sofia Lavieri We derive optimal cholesterol treatment plans for patients at risk of MD07 MD08

cardiovascular events due to clinical and genetic factors using a Markov decision process. We also introduce a simulation-based framework to estimate the risk of a coronary heart disease event due to genetic factors. 2 - Optimal Inventory Control for Adaptive Trial Supply Wei-An Chen, Purdue University, 610 Purdue Mall, West Lafayette, IN, United States, chen1702@purdue.edu, Nan Kong As adaptive clinical trials receive growing attention in the past decade, a clinical supply management tool is required to address the adaptiveness. Our research aims at bringing more flexibility to drug supply and coordinate it with adaptive trial design. Our stochastic optimization model is founded on drug inventory control and used to address several important aspects of drug supply, e.g., drug wastage, resupply policy, trial span, and overall cost, under the uncertainty of participant enrollment. With the focus of trial size and length re-estimation, we also consider jointly optimize the trial design parameters and supply decisions while incorporating the patient response uncertainty. 3 - Optimal Risk-based Group Testing Hrayer Y Aprahamian, Virginia Tech, 1145 Perry Street, Blacksburg, VA, 24060, United States, ahrayer@vt.edu Group testing, i.e., testing multiple subjects simultaneously within a group, is an essential tool for classifying subjects as positive or negative for a binary characteristic; and is utilized in a variety of settings. We study optimal group testing schemes that incorporate subject-specific risk characteristics into the testing design. We consider various objectives, and characterize key structural properties of optimal testing designs. These properties allow us to develop efficient algorithms and derive important insights on the trade-offs being incurred. Our case study considers Chlamydia screening, and demonstrates the Kalyan Pasupathy, Mayo Clinic, Rochester, MN, 4Mayo Clinic, Rochester, MN, Contact: pasupathy.kalyan@mayo.edu, Ozden Onur Dalgic, Fatih S. Erenay, Osman Yalin Ozaltin, Mustafa Y. Sir, Brian A. Crum ALS is a neurodegenerative disease causing continuous decay of motor neurons and muscle atrophy. Patients suffer from losing their abilities to speak, eat, move and breath. Due to having no permanent treatment, ALS eventually affects all abilities but disease progression shows a great variability. Using medical records of over 400 patients from Mayo Clinic, we analyze the ALS progression pathways, and estimate risks of losing abilities and needing medical interventions (e.g., feeding tube, ventilator). We then develop a stochastic natural history model to optimize the timing of ALS interventions for improving patients’ well-being in their disease course. Yun Fong Lim, Associate Professor, Singapore Management University, 50 Stamford Road, 04-001, Singapore, 178899, Singapore, yflim@smu.edu.sg, Song Jiu, Marcus Teck Meng Ang We consider an online retailer selling multiple products over a multi-period setting. At the start of each period, the retailer replenishes her products from multiple suppliers. After receiving the products, the retailer allocates the inventory to different fulfillment centers. At the end of each period when the demands for the products from different zones are realized, the retailer fulfills them from different fulfillment centers. We formulate a robust optimization model to determine the replenishment and allocation decisions at the start of each period and the fulfillment decisions at the end of the period. Our approach can handle large-scale problem instances with good quality solutions. 2 - A Study of Distributionally Robust Multistage Stochastic Optimization Yongpei Guan, University of Florida, 303 Weil Hall, P.O. Box 116595, Gainesville, FL, 32611, United States, guan@ise.ufl.edu, Kezhou Zhou, Jianqiu Huang In this talk, we focus on a data-driven risk-averse multistage stochastic programming (RMSP) model considering distributional robustness. We optimize the RMSP over the worst-case distribution within an ambiguity set of probability distributions constructed directly from historical data samples. The proposed RMSP is intractable due to the multistage nested minimax structure in its objective function, so we reformulate it into a deterministic equivalent that contains a series of convex combination of expectation and conditional value at risk (CVaR), which can be solved by a customized stochastic dual dynamic programming (SDDP) algorithm. As the size of collected data samples increases to infinity, we show the consistency of the RMSP with distributional robustness to the traditional multistage stochastic programming. value of risk-based testing within a public health framework. 4 - Optimal Timing of ALS Interventions to Improve the Disease Course MD09 330A Robust Optimization and its Application Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Daniel Long, zylong@se.cuhk.edu.hk 1 - Matching Supply with Demand for Online Retailing

238