Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WC59

2 - Resource Allocation for Medical Fraud Assessment Tahir Ekin, Texas State University, San Marcos, TX, 78666, United States, Rasim Musal It is estimated that up to 10 percent of annual health care spending is lost to fraudulent transactions, which makes medical audits paramount. The resource allocation decisions for the medical audits are challenging because of the heterogeneity of the claims. A number of frameworks are utilized to help auditors address the trade-offs between efficiency and cost while having valid overpayment amount estimates. We present resource allocation designs to help the auditors assess the trade-off between the initial and additional numbers of samples within a multistage sampling framework. 3 - Productivity-driven Physician Scheduling in Emergency Departments Nadia Lahrichi, Ecole Polytechnique de Montreal, Cp6079 Succ. Centre-Ville, Montreal, QC, H3C3A7, Canada, Dina Bentayeb, Louis-Martin Rousseau Reducing the congestion in EDs is a major challenge for improving access to care. Our approach to deal with this problem is to offer an innovative approach to scheduling emergency physicians. Currently, departments define a covering constraint that fixes the number of required physicians per shift/day. We observe two main issues: 1) the variation in demand in terms of number of patients is not considered when designing the schedule; 2) all physicians are considered equivalent in terms of the number of patients they see in a shift. These two aspects lead to a mismatch between offer and demand. In this project, we will address these two issues to better align the schedule of physicians to patients demand. 4 - Patient Prioritization in an Ambulatory Care Clinic Using Markov Chain Models Vahab Vahdatzad, Northeastern University, Boston, MA, 02114, United States, Jacqueline Griffin, Ann Suhaimi In outpatient centers, physicians regularly visit patients with different characteristics. Furthermore, physicians may visit each patient multiple times during one visit. An example of multiple visit is in orthopedic centers that physician visit the patient before and after radiology tests. In this research, prioritizing patients are examined using Queuing Network and Markov chain models and showed how altering prioritization based on patient characteristics can improve timeliness of care for all patients. n WC59 West Bldg 102A Operational Decisions in Healthcare Sponsored: Health Applications Sponsored Session Chair: Zhankun Sun, City University of Hong Kong, Kowloon, 1, Hong Kong 1 - Design of Specialist Response Policies to Reduce Waiting Time in Emergency Departments Cheng Zhu, McGill University, 701-801 Sherbrooke Est, Montreal, QC, H2L 0B7, Canada, Beste Kucukyazici We design a framework of systematic response policies for specialists based on their clinical demands. After identifying a class of patients who are more likely to require specialist consultation based on their clinical information available at the triage stage through statistical analysis, we analyze several alternative policies for specialists’ responses to consultation requests using queueing models. Specifically, we analytically calibrate a specialist’s optimal arrival in a cycle based on the non- homogenous Poisson distributed demand. The simulation suggests this policy can be effectively adopted in different types of hospitals with varied patient volumes. 2 - Bed Allocation to Reduce Overflow Jingui Xie, University of Science and Technology of China, School To address the overflow issue, we build an analytical model and propose two easy-to-compute bed allocation policies. We use the real data from the only university hospital in Singapore and a simulation model to evaluate the effectiveness of our proposed policies against the base case provided by the empirical study of the hospital. Through the simulation study, we show that the proposed policies can reduce the overflow rate from 18.91% to about 4-5% without sacrificing other performance measures. More surprisingly, our simulation studies suggest that the existing capacity actually can accommodate 50% more elective patients while keeping the overflow rate at a level of less than 10%. of Management, 96 Jinzhai Road, Hefei, 230026, China, Marcus Teck Meng Ang, Mabel Chou, David D. Yao

n WC57 West Bldg 101B Mathematical Modeling and Optimization for Cancer Therapy Sponsored: Health Applications Sponsored Session Chair: Ehsan Salari, Wichita State University, Wichita, KS, 67260, United States 1 - Dose Mimicking with Inverse Optimization Aaron Babier, University of Toronto, Toronto, ON, M5S 2K6, Canada, Andrea McNiven, Timothy Chan There are many instances in radiotherapy (e.g., knowledge-based planning) where we know what a desirable dose distribution looks like, but we do not know how to deliver it to a patient. In these cases, dose mimicking methods are used to optimize the beamlets that best deliver the plan in question. We propose an inverse optimization (IO) pipeline as an improvement over conventional dose mimicking methods. After testing both approaches on clinical plans and predictions from a random forest, we found that our IO approach can mimick plans better than a conventional method. These results were consistent across a large cohort of 217 patients, and suggest that IO is a viable tool for dose mimicking. 2 - Effects of Spatial Structure on the Speed of Evolution Nathaniel Witte, University of Minnesota, Minneapolis, MN, United States, Kevin Leder Spatially structured and spatially homogeneous models are both commonly used to model cancer growth. In this work, we consider two stochastic models of tumor evolution: a logistic branching process and an interacting particle systems. In particular, we are interested in how the average population fitness changes versus time in the two models. We also investigate how spatial structure affects tumor response to therapy. 3 - Optimizing Spatiotemporally Fractionated Radiotherapy Plans Ali Adibi, Wichita State University, Wichita, KS, 67220, United States, Ehsan Salari Several studies have recently shown the potential therapeutic gain that may be achieved from delivering spatiotemporally fractionated radiotherapy plans. However, solving the corresponding treatment planning problem to a small optimality gap remains computationally challenging for 3D clinical cases due to the large-scale and non-convex nature of the problem. This research aims at developing a customized solution approach to spatiotemporal fractionation. 4 - Organ Motion Prediction in MRI Guided Radiotherapy Seyed Ali Mirzapour, Wichita State University, Wichita, KS, 67220- 2982, United States, Thomas Mazur, Gregory Sharp, Ehsan Salari Internal organ motion during radiation therapy (RT) is a major concern in the treatment of lung and abdominal cancers. If unaccounted for, the anatomical motion may lead to the underdosing of the target volume or overdosing of organs at risk. A new generation of RT systems are equipped with onboard magnetic- resonance imaging (MRI) scanners, providing a real-time view of the anatomical motion during RT delivery. This research aims at employing the online stream of MRI images to predict the motion trajectory using Markov processes. n WC58 West Bldg 101C Resource Allocation in Health Care Sponsored: Health Applications Sponsored Session Chair: Jackie Griffin, Northeastern University, Boston, MA, 02115, United States 1 - Approximate Dynamic Programming for the Aeromedical Evacuation Dispatching Problem: Value Function Approximation Using Multiple Level Aggregation Nathaniel D. Bastian, Assistant Professor, Department of Systems Engineering, U.S. Military Academy, West Point, NY, 10996, United States, Matthew Robbins, Phillip Jenkins, Brian Lunday Sequential resource allocation decision-making for the military medical evacuation of wartime casualties consists of identifying which available aeromedical evacuation (MEDEVAC) assets to dispatch in response to each casualty event. These sequential decisions are complicated due to uncertainty in casualty demand and service times. In this research, we present a Markov decision process model solved using a hierarchical aggregation value function approximation scheme within an approximate policy iteration algorithmic framework. The model seeks to optimize this sequential resource allocation decision under uncertainty of how to best dispatch MEDEVAC assets to calls for service.

489