2016 INFORMS Annual Meeting Program

MC23

INFORMS Nashville – 2016

2 - Supporting The Decision Making Process In Safety Review Taxiarchis Botsis, FDA/CBER/OBE, Taxiarchis.Botsis@fda.hhs.gov Medical experts at the US Food and Drug Administration (FDA) conduct surveillance of licensed medical products to assure continued safety. The review process requires the thorough evaluation of multiple parameters and can be time- consuming for the end users. To assist them in this process, the FDA has developed a Decision Support Environment that extracts key clinical (and time) information from the texts, normalizes it to medical codes, and visualizes it in meaningful manners. It also allows for other analysis, case management and report generation. 3 - The Impact Of Transfusing Newer Blood Versus Current Practice On The US Blood Supply Hussein Ezzeldin, FDA/CBER/OBE, Silver Spring, MD, United States, hussein.ezzeldin@fda.hhs.gov, Richard Forshee, Arianna Simonetti The prolonged storage of red blood cells (RBCs) may be associated with transfusion adverse events. Mixed findings between observational and clinical studies demand larger and better-designed studies. We enhanced the US blood supply model by implementing an adaptive feedback-control mechanism, where blood collection is dynamically adjusted, to maintain inventory collector’s levels and overcome potential shocks. If any benefits in transfusion outcomes of younger RBCs are proven, an increase in their demand may be expected. We evaluate the impact of such changes on the US blood system with respect to current blood transfusion practice. 4 - Optimal Resource Allocation For Adaptive Clinical Trials Alba Rojas-Cordova, Virginia Tech, Blacksburg, VA, 24060, United States, albarc@vt.edu, Ebru Korular Bish Certain adaptive designs allow decision-makers to alter the course of a clinical trial based on revised estimates of a drug’s probability of technical success. We develop a stochastic dynamic programming model to analyze the resource allocation decision, of continuing or stopping a trial, based on frequent data updates. We determine the structure of the optimal resource allocation strategy and support our findings with a numerical analysis. MC22 107B-MCC Cost, Safety and Resource Allocation In Health Systems Invited: ORinformed Healthcare Policies Invited Session Chair: Retsef Levi, MIT, Cambridge, MA, United States, retsef@mit.edu 1 - Data-driven Approaches To Improve The U.S. Kidney Allocation System We present a data-driven optimization approach to estimate wait times for individual patients in the U.S. Kidney Allocation System, based on the very limited system information that they possess in practice. To deal with this information incompleteness, we develop a novel robust optimization analytical framework for wait time estimation in multiclass, multiserver queuing systems. We calibrate our model with highly detailed historical data and illustrate how it can be used to inform medical decision making and improve patient welfare. 2 - Optimization-driven Framework To Understand Healthcare Networks Cost And Resource Allocation Fernanda Bravo, UCLA Anderson School of Management, Los Angeles, CA, United States, fernanda.bravo@anderson.ucla.edu Marcus Braun, Vivek Farias, Retsef Levi Consolidation in the US healthcare industry has resulted in the formation of large delivery networks. However, integration remains uncertain. In order for large care providers to best utilize their growing networks, it will be critical to understand not only system-wide demand and capacity, but also how the deployment of limited resources can be improved. We develop an optimization- driven framework, inspired by revenue management, to understand network costs and provide solutions to strategic problems, such as access, resource deployment, and case-mix in multi-site networks. In collaboration with a network of hospitals, we demonstrate our framework applicability. Nikolaos Trichakis, MIT, Cambridge, MA, United States, ntrichakis@mit.edu, Chaithanya Bandi, Phebe Vayanos

3 - New Data-driven Approach To Safety And Risk Management In Health Systems Retsef Levi, Professor of Operations Management, MIT, 100 Main Street, BDG E62-562, Cambridge, MA, 02142, United States, retsef@mit.edu, Patricia Folcarelli, Yiqun Hu, Daniel Talmor, Jeffrey Adam Traina We present an innovative system approach to safety in Health Systems. The approach is based on the innovative concept of risk drivers, which are states of the System, its environment and its staff that affect the likelihood of harms, as well as an innovative aggregated measure of the ‘burden of harm’. Using large scale data we develop statistical models that identify predictive risky states. 108-MCC Healthcare Delivery Modeling Sponsored: Health Applications Sponsored Session Chair: Bryan A Norman, University of Pittsburgh, 1006 Benedum Hall, Pittsburgh, PA, 15261, United States, banorman@pitt.edu 1 - Modeling To Enhance The Nurse Handoff Process Anna Svirsko, University of Pittsburgh, ACS167@pitt.edu Bryan A Norman, David Rausch, Emily Shawley Nurses in emergency departments often rotate between different zones during their 12 hour shift to balance nurse workload. However, this rotation significantly increases the number of nurse handoffs which reduces the amount of available nurse-patient time and can result in errors in patient care. This model reduces the number of nurse handoffs while still allowing nurses to rotate during their shift and balancing workload among nurses. Furthermore, we look to reduce long chains and cycles that can occur that hinder the rotation and handoff processes. The effectiveness of the model is demonstrated by applying the model to the nursing schedule from a local hospital. 2 - Closed-Form Solutions For Periodic Review Inventory Systemsin Healthcare Nazanin Esmaili, University of Pittsburgh, Pittsburgh, PA, United States, nae22@pitt.edu, Bryan A Norman, Jayant Rajgopal Most inventory management systems at points of use in hospitals are characterized by stochastic demand, periodic reviews, fractional lead time, expedited delivery, limited storage capacity, and service level requirements. We develop discrete time Markov chain models for such systems to minimize the total expected replenishment effort. We derive closed form expressions and propose an exact algorithm to calculate the limiting probability distribution by locally decomposing the state space. We investigate the structural results and the tradeoffs of performance measures of interest for different policies and show that the computational effort is less than other algorithms from the literature. 3 - Considering No-shows And Procedure Time Variability When Scheduling Endoscopy Patients Karmel Shehadeh, University of Michigan, Ann Arbor, MI, United States, ksheha@umich.edu, Amy Cohn, Sameer Saini, Jacob Kurlander We consider the problem of how to schedule patients for colonoscopy appointments, recognizing both the high frequency of no-shows and the significant variability in procedure time. We review the clinic process flow, identify metrics for evaluating schedule quality, and simulate different scheduling approaches. 4 - Improving Healthcare Resource Management Through Demand Prediction And Staff Scheduling Nazanin Zinouri, Clemson University, 269 Freeman Hall, Clemson, SC, 29634, United States, nzinour@g.clemson.edu, Kevin Taffe Staff scheduling in healthcare is very challenging. Hospitals typically operate 24 hours a day, 7 days a week, and are faced with high fluctuations in demand. We developed an ARIMA model to forecast daily patient volumes a month in advance. This information was used to compute workload and solve staff scheduling problems. We used a risk adverse approach to find a feasible nurse assignment that minimizes labor costs and to avoid risky cases, i.e., highly overstaffed or understaffed. The liabilities of overstaffing and understaffing are many. Overstaffing increases payrolls and results in excessive idle times, while understaffing will negatively impact patient outcomes and results in loss of revenue. MC23

189