Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WD61

High variability of surgery durations is a major challenge towards creating practical OR schedules. To address this issue, we proposed a two-stage stochastic model with chance constraints. The first stage makes decisions on OR openings and case assignments to minimize fixed costs. The second stage aims to minimize the expected penalty costs after observing the surgery durations. Chance constraints control the risk of having OR overtimes. A decomposition-based method is developed based on the structure of the model. Coefficient strengthening and bounding are performed to accelerate the solution process. Numerical results show that our algorithm can handle large instances in a timely manner. 2 - Online Scheduling for Outpatients Kimia Ghobadi, MIT, 100 Main Street, E62-459, Cambridge, MA, 02142, United States, Retsef Levi, Michael Hu In this talk, we demonstrate the use of real-time algorithms to improve patient care. We discuss resource utilization in infusion clinics. Many outpatient clinics, including infusion clinics, struggle to meet their demand due to crippling congestion in peak hours. At the same time, these clinics are often underutilized at non-peak hours. This underutilization is induced primarily by inappropriate scheduling practices. Therefore, we introduce a new scheduling model and employ real-time algorithms to improve efficiency by allowing the clinics to treat more patients with fewer resources. We show 30% empirical improvement and theoretical worst-case bounds on the performance of the algorithm. 3 - Inferring Objective Functions from Inconsistent Data in Healthcare and Energy Applications Taewoo Lee, University of Houston, E209 Engineering Bldg 2, 4722 Calhoun Rd, Houston, TX, 77204-4008, United States, Zahed Shahmoradi Given often inconsistent observations as input data, we develop a new inverse optimization model that determines a set of objective functions of a linear program that render the most relevant subset of the observations near-optimal. We analyze the feasibility and optimality of the nonconvex inverse model and propose an algorithm that finds all inverse-feasible objective functions. Our approach addresses the infeasibility and instability issues of the previous inverse models. We demonstrate the model in the context of multi-objective diet recommendation and electricity demand forecasting. 4 - Data-driven Objective Selection in Multi-objective Optimization Temitayo Ajayi, Rice University, Houston, TX, 77025, United States, Taewoo Lee, Andrew J. Schaefer A challenge in radiation therapy treatment planning is selecting which clinical objectives to use in the optimization. We propose an inverse optimization method with a cardinality constraint to infer the most important objectives from historical treatment plans. We use a greedy algorithm to select objectives and provide theory, a generalization of a result by Nemhauser (1978), to support our results. We compare the proposed method to the cardinality-constrained inverse problem and show that our method efficiently finds a small number of objectives that generates clinically acceptable treatment plans. n WD61 West Bldg 102C Practice – Supply Chain Competition, Practice and Risk Analysis Contributed Session Chair: Jian Li, Northeastern Illinois University, 5500 N. St Louis Avenue, Chicago, IL, 60625, United States 1 - Supplier Development and Learning Spillover in a Triadic Supply Chain under Competition Abhishek Srivastava, Doctoral Student, Indian Institute of Management Kozhikode, IIMK PO Kunnamangalam, Kozhikode, 673570, India, Arqum Mateen We model a triadic supply chain in which two competing asymmetric buyers source from common supplier, invest in supplier development to improve profits. Buyers may benefit from each other’s investment equally (uniform spillover; US) or unequally (differential spillover; DS). The supplier offers a uniform wholesale price (UWP) and buyer-specific wholesale price (BWSP). Our results indicate that optimal wholesale price increases with market share, intensity of competition, spillover effect. UWP under US works against the larger buyer, who consequently prefers a DS structure. We establish that both the buyers prefer investment under DS structure for higher values of competition and spillover. 2 - A Study on Logistics Companies’ Financial and Supply Chain Performance Hossein Najmi, University of North Texas, 1307 W. Highland Street, Denton, TX, 76201, United States, Seock Hong Financial performance indicators (FPIs) are long used to find supply chain excellence. But examining FPIs specific to a sector is understudied. To address this gap, we applied discriminant analysis, logistic regression, and MANOVA to analyze 103 logistics companies. We found the FPIs that measure the performance of logistics companies supply chain and differences among industries.

n WD55 North Bldg 232C Practice- Technology and Project Management Contributed Session Chair: Christian Ruf, Technical University Munich, Arcisstr. 21, München, D-80333, Germany 1 - Motivating Employees’ Information Security Compliance: Leadership Style or Protection Motivation? Jiawen Zhu, Xi’an Jiaotong University, Xi’an, China Jiawen Zhu, City University of Hong Kong, Hong Kong, China, Gengzhong Feng, Kwok Leung Tsui This study investigates the moderation effect of Paternalistic Leadership on the relationship between Protection Motivation Theory and employeesæ Information Security Compliance behavior. Using questionnaire data from 760 employees, we found that, Paternalistic Leadership dimensions, Authoritarianism, Benevolence, Morality, dampen the weight of most protection motivation elements to employees’ compliance. This suggests that when certain leadership style is salient, employees’ protection motivation perceptions, such as threat severity, self-efficacy and maladaptive reward perception, may have no significant influence on employees’ information security compliance. 2 - Asymmetric Information Sharing in Information System Security Yueran Zhuo, University of Massachusetts Amherst, 121 Presidents Drive, Amherst, MA, 01003, United States, Senay Solak In information security practice, the asymmetry in information sharing levels might hurt the information sharing firms’ incentives. A possible solution to the problem is to impose charges on the shared information and treat it as a commodity. In this study we try to answer the questions: What fair price should a firm pay when also sharing a certain level of information? How would the price of information vary as more firms join an information sharing alliance? We develop analytical expressions to identify the pricing of information in an information sharing community with multiple firms and explore the overall benefits to the information sharing community due the implementation of pricing strategies. 3 - Introducing a Predator-prey System with Michaelis-menten Type of Prey and Predator Harvesting Considering Diffusion Terms and Inconstant Carrying Capacity for the Prey Aram Bahrini, University of Virginia, Thornton Hall, P.O. Box 400259, Charlottesville, VA, 22904-4259, United States University of British Columbia, 1984 Mathematics Road, UBC, Vancouver, BC, V6T1Z2, Canada, Behnam Malmir, Mohammad Najjartabar Bisheh In this paper, we made a modification on the Hu and Cao (2017) model in which in our model the effect of harvesting for the prey is non-zero and considered as a Holling II functional response predator-prey system. In addition, since by changing the reproduction rate, the carrying capacity may change as well, the scenario in which the carrying capacity is considered as unfixed and a function of reproduction rate for the pray is taken into consideration. Additionally, a diffusion term is added to make the model more realistic especially in fishery application. Results show that the model analyzed in this paper can cause more realistic behaviors compared with other presented systems. 4 - Long-term Capacity Planning of a Workforce with Hierarchical Skills and Random Resignations Christian Ruf, TU München, Arcisstr. 33, Munich, 80333, Germany, Jonathan F. Bard, Rainer Kolisch We address a multistage capacity planning problem for a hierarchically skilled workforce. Recruits are hired and trained over multiple periods to perform jobs that require ever greater skills. The training comprises a combination of off-the- job and on-the-job elements. Random resignations result in labor shortfalls that jeopardize continuous operations. The problem is modeled as a Markov decision process for which a parameterized decision rule is proposed. To determine good parameter values, we present a very large-scale neighborhood search. Experimental results on real-world industry data is presented. n WD56 West Bldg 101A Data-driven Modeling in Healthcare Sponsored: Health Applications Sponsored Session Chair: Taewoo Lee, University of Houston, Houston, TX, 77204-4008, United States 1 - A Two-stage Stochastic Model for the Operating Room Scheduling Problem with Chance Constraints Amirhossein Najjarbashi, University of Houston, E206 Engineering Bldg 2, 4722 Calhoun Rd, Houston, TX, 77204, United States, Gino J. Lim

513