Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TB59

2 - Dynamic Appointment Scheduling of Elective Surgeries Under Bed Capacity Constraint

n TB59 West Bldg 102A Joint Session HAS/Practice Curated: Applications of Decision-Making Models in Healthcare Sponsored: Health Applications Sponsored Session Chair: Alireza Boloori, Arizona State University, Tempe, AZ, 85283, United States 1 - Bias in Sensitivity Analysis of Comparative Analyses for Medical Decision Making Michael J. Hintlian, University of Southern California, 1029 South Westmoreland Avenue, #102, Los Angeles, CA, 90006, United States, Julia L. Higle Comparative analyses for MDM are undertaken examine the cost/benefit impact of various treatment alternatives. These impacts are estimated via model-based analyses after which sensitivity to model parameters is examined. We illustrate the existence of bias in the sensitivity analysis that results from the methods used to select model parameters. We discuss methods for mitigating this bias. 2 - Optimal Genetic Testing Schemes for Cystic Fibrosis Hussein El Hajj, Virginia Tech, Blacksburg, VA, United States, Ebru Korular Bish, Douglas R. Bish Cystic fibrosis (CF) is a highly prevalent life-threatening genetic disorder, but early diagnosis can save lives and reduce healthcare expenditures. To date, over 300 CF-causing mutations are identified, and all 50 states conduct newborn screening for CF, typically starting with a bio-marker test, followed by genetic testing on selected mutations for newborns with elevated bio-marker levels. We develop a stochastic optimization model to determine an optimal genetic testing scheme for CF that minimizes the probability of misclassification under a testing budget. Our case study for California shows that the optimal scheme can substantially reduce misclassification over current practices. 3 - Incentive-Driven Readmission Management with Patients Facing Compliance Barriers Aditya Mahadev Prakash, University of Florida, Gainesville, FL, 32608, United States, Qiaochu He, Xiang Zhong We aim to quantify the impact of non-compliance of patients on their post-acute care management, and assist healthcare stakeholders in improving the overall well-being of patients through the most efficient and effective allocation of resources. We establish a game-theoretic model where patients’ lack of compliance is modeled by incorporating their heterogeneous and bounded rationality in the context of a congested service system. The optimal structure of subsidies that can monetarily incentivize patients and result in a minimum overall cost for an insurer is developed. The insights obtained from this study would support clinical and operational decision-making by health practitioners. 4 - Impact of Physician’s Ambiguity on Management of Medications Alireza Boloori, Arizona State University, Tempe, AZ, 85283, United States, Soroush Saghafian, Harini A. Chakkera, Curtiss B. Cook Patients after organ transplantations receive high amounts of immunosuppressive drugs (e.g., tacrolimus) to reduce the risk of organ rejection. However, this practice has been shown to increase the risk of New-Onset Diabetes After Transplantation (NODAT). We propose an ambiguous POMDP framework to generate effective medication management strategies for tacrolimus and insulin. Our approach increases the patient’s quality of life while reducing the effect of transition probability estimation errors. We also provide several managerial and medical implications for policy makers and physicians. n TB60 West Bldg 102B Scheduling in Healthcare Operations Sponsored: Health Applications Sponsored Session Chair: Jin Qi, Hong Kong University of Science and Technology Hong Kong 1 - A Predictive Analytics Framework for Personalizing Patient Encounters for Diabetes Care Han Ye, U. of Illinois at Urbana-Champaign, 350 Wohlers Hall, 1206 South Sixth Street, Champaign, IL, 61820, United States, Ujjal Kumar Mukherjee, Dilip Chhajed In this paper, we develop a two stage decision framework that considers predicted diabetes risks of individual patients as inputs, which healthcare organizations can implement for optimally scheduling patient encounters for management of diabetes care.

Chengyu Wu, Duke University, Durham, NC, 27705, United States, Li Chen, Jing-Sheng Jeannette Song

Both elective surgery patients and emergency surgery patients, after leaving operating room, need to enter the ICU that has a limited bed capacity. We study the problem of optimally scheduling elective surgeries in advance by taking into account the ICU capacity and occupancy. In doing so, we derive structural results and develop a heuristic which is validated using real-world hospital data. 3 - Appointment Scheduling with a Waiting Time Target Xing Liu, City University of Hong Kong, CITYU, Hong Kong, Frank Y. Chen, Jin Qi, Han Zhu This work is motivated by the appointment booking of a care center, which accepts only advanced booking and patients should receive consultations within a stipulated target waiting time. We propose a heuristic for this advance booking problem through the policies of an allocation scheduling counterpart of the problem. 4 - Operating Theater Scheduling Under Uncertainty with an Entropic Index Xiaojin Fu, Hong Kong University of Science and Technology, Hong Kong, Jin Qi, Han Ye We consider a surgery scheduling problem in an operating theater with uncertain surgery durations. We introduce the Entropic Tardiness Index (ETI) to quantify both the frequency and intensity of surgery delay or OR overtime. A mathematical model is formulated to find a sequencing decision which minimizes the ETI criterion, and an algorithm based on benders decomposition is developed to find the optimal solution. Inspired by a heuristic, we propose an index Entropic Deviation (ED) to account for both variation and skewness of surgery durations. Numerical study shows that sequencing decision obtained by sorting the EDs of the surgeries in the ascending order achieves a relatively good performance. n TB61 West Bldg 102C Optimization Society Award Session I Sponsored: Optimization Sponsored Session Chair: David Morton, Northwestern University, IEMS Department, 2145 Sheridan Road, Evanston, IL, 60208, United States 1 - A Brief Tour of Logic and Optimization John Hooker, Carnegie Mellon University, Tepper School of Business, Pittsburgh, PA, 15213, United States This talk conducts a brief tour of the remarkable connections between logic and optimization, some of which have led to substantial improvements in solution technology. They include the fundamental role of logic in optimization duality and how it leads to logic-based Benders decomposition; the deep relationship between cutting planes and logical inference methods; linear and integer programming models for probabilistic, nonmonotonic, and belief logics; the role of logic’s famous Herbrand Theorem in infinite-dimensional integer programming; binary decision diagrams as a discrete optimization tool; and how concepts of consistency in constraint programming relate to cutting plane theory. 2 - A Simple Nearly-Optimal Restart Scheme for First-Order Methods James Renegar, Cornell University, Ithaca, NY, United States I was lucky as a researcher to be coming of age when the “interior-point method revolution” was beginning to unfold, lucky to be in position to make a research contribution that was quickly and widely appreciated, a contribution that has played a significant role in now being honored with the Khachiyan Prize. The talk, however, focuses on recent research, presenting a simple restart scheme for first-order methods, which for rich classes of convex optimization problems results in algorithms with complexity bounds nearly matching lower bounds established by Nemirovski and Nesterov in the 1980’s. 3 - Stochastic Programming: Approximations and Scenarios Werner Roemisch, Humboldt Universitat, Berlin, Germany Solving a stochastic program requires approximating a probability distribution with a finite set of scenarios. First, we review scenario generation using stability and probability metrics. Second, we describe a stability- and problem-based method to generate scenarios. For two-stage SPs this requires solving a generalized semi-infinite program, which reduces to a standard convex semi- infinite program in special cases. Third, we review quadrature using Monte Carlo, Quasi-Monte Carlo and sparse grid, and the method of moments, including convergence properties. We discuss numerical results.

296