2016 INFORMS Annual Meeting Program

TB24

INFORMS Nashville – 2016

TB25 110A-MCC Scheduling and Contracts Invited: Project Management and Scheduling Invited Session Chair: Nicholas G Hall, Ohio State University, Columbus, OH, United States, hall.33@osu.edu 1 - Multitasking Via Alternate And Shared Processing: Algorithms And Complexity Chung-Lun Li, The Hong Kong Polytechnic University, chung-lun.li@polyu.edu.hk, Nicholas G Hall, Joseph Leung This work is motivated by disruptions that occur when jobs are processed by humans, rather than by machines. E.g., humans may become tired, bored, or distracted. We present two scheduling models with multitasking features, which aim to mitigate the loss of productivity in such situations. The first model applies “alternate period processing” and aims either to allow workers to take breaks or to increase workers’ job variety. The second model applies “shared processing” and aims to allow workers to share a fixed portion of their processing capacities between their primary tasks and routine activities. For each model, we consider several widely studied and practical classical scheduling objectives. 2 - Scheduling To Minimize Energy Cost Marc Posner, The Ohio State University, posner.1@osu.edu, Nicholas G Hall While scheduling is an effective way to improve energy efficiency in manufacturing, optimal scheduling becomes more complicated when energy costs vary. Our machine has discretely variable speeds, and increased energy usage is incurred at faster machine speeds. Three alternative scenarios about the time at which the machine can change speed are considered. In each scenario, we study the problem of minimizing total energy cost, subject to the completion of work by a given date. We describe efficient algorithms for these problems where possible, and also identify limits to their solvability. 3 - Analysis Of A Procurement Game With Option Contracts Bo Chen, University of Warwick, Coventry, United Kingdom, b.chen@warwick.ac.uk, Edward James Anderson, Lusheng Shao When a firm faces an uncertain demand, it is common to procure supply using some type of option in addition to spot purchases. A typical version of this problem involves capacity being purchased in advance, with a separate payment made that applies only to the part of the capacity that is needed. We address such a problem by formulating it as a procurement game, in which competing suppliers choose a reservation price and an execution price for blocks of capacity, and the buyer, facing known distributions of demand and spot price, needs to decide which blocks to reserve. 4 - Scheduling Crash Tests At Ford With Sequencing Restrictions And Capacity Constraints Yuhui Shi, University of Michigan, 2212 Glencoe Hills Drive, Apt 11, Ann Arbor, MI, 48108, United States, yuhuishi@umich.edu, Amy Cohn, Marina Alex Epelman We present the problem of scheduling crash tests at Ford Motor for new vehicle model development. In this problem, we assume performing crash tests requires prototype vehicles as well as other limited supporting resources such as testing facilities and engineers. We show how to solve the problem subject to these resource constraints by using various decomposition methods. Invited: Auctions Invited Session Chair: Benjamin Lubin, Boston University, Boston, MA, United States, blubin@bu.edu 1 - Adaptive-price Combinatorial Auctions Benjamin Lubin, Boston University, blubin@bu.edu This work introduces and implements an iterative combinatorial auction that aims to achieve both high efficiency and fast convergence without prior restrictions on the valuation domain. Our auction uses polynomial prices, which price combinations of items beyond just single items, and gradually extends price expressivity as the rounds progress. We also propose a heuristic approach to winner determination to ensure the auction scales. An experimental evaluation shows that our auction is competitive with bundle-price auctions in regimes where these excel, namely multi-minded valuations, but also performs well in regimes favorable to linear prices, such as valuations with pairwise synergy. TB26 110B-MCC Combinatorial Auction Pricing

4 - A Continuous Time Stochastic Model To Optimize Blood Pressure Treatment Decisions Anthony Bonifonte, Georgia Institute of Technology, ABonifon@gatech.edu, Turgay Ayer, Ben Haaland, Peter Wilson Antihypertensive drug treatment can control elevated blood pressure and reduce the risk of future cardiovascular outcomes. We develop a data-driven stochastic model of blood pressure progression that generalizes Brownian motion by modeling the change in blood pressure per unit time as a Gaussian mixture distribution. This model addresses the question of what thresholds at which to initiate antihypertensive treatment and the optimal intensity. Our main finding is initiation and intensity decisions depend jointly on systolic and diastolic pressure. The methods are generalizable to other chronic diseases with continuous valued measurements. 5 - Designing Effective Vaccine Administration Practices Gizem Sultan Nemutlu, PhD Candidate, University of Waterloo, Faculty of Engineering University of Waterloo, Carl A. Pollock Hall 3654 200 University Avenue West, Waterloo, ON, N2L 3G1, Canada, gsnemutlu@uwaterloo.ca, Fatih Safa Erenay, Osman Yalin Ozaltin Childhood vaccine wastage due to limited shelf-life of opened vials is still high in developing countries. Our research shows that open vial wastage can be significantly reduced by keeping vaccine stocks in different size vials, and dynamically deciding what size of vial to use next and/or when to terminate daily vaccination services. We develop a discrete-time MDP model maximizing demand coverage. We analyze the structural properties of the optimal strategies and show that the proposed model can help decision makers in determining the best vial- size combinations and optimal inventory levels. Chair: John Zaleski, Bernoulli Health, 4801 S. Board Street, Suite 120, Philadelphia, PA, 19112, United States, jzaleski@bernoullihealth.com 1 - Clinical Applications Of Data Analytics: A Survey Amy Harris, Middle Tennessee State University, amy.harris@mtsu.edu Increasing volumes and varieties of clinical data and growing interest in analytics has created opportunities for healthcare organizations to study and address clinical problems. Experience gained is often employed to improve care and, where successful, results are published in academic journals. This presentation surveys the literature and explores which types of analytics are most popular and how healthcare organizations use data analytics to inform clinical practice. 2 - The Use Of Kalman Filtering In Alarm Management Studies John Zaleski, Bernoulli Enterprise Inc, jzaleski@bernoullihealth.com Vitals signs monitoring in high acuity environments are often the source of alarms and notifications to care providers. In this presentation, the author demonstrates the use of a Kalman filtering technique that has been used in the identification of alarm limit thresholds for capnography monitoring of patients in the medical surgical environment post-operatively. 3 - The State Of Digital Marketing In The Healthcare Industry Brian Harris, Lima Consulting, bharris@limaconsulting.com Digital marketing technologies and web analytics have opened new opportunities to provide actionable analysis to decision makers. However, many analysts struggle with inconsistent or questionable data due to failed or sub optimal deployments of these technologies. This study examines two million pages across websites of hundreds of the world’s largest healthcare companies to provide insights on the relative health of digital marketing and web analytics across multiple sectors. It also provides practical advice for analysts seeking to improve the quality and veracity of their web analytics data. 4 - Consistent Staffing For Long-term Care Through On-call Pools Nursing home managers have increasingly emphasized consistency of care — i.e., minimizing the number of different nurse aides who care for each resident — but have struggled with this goal due to nurse aide absences before the start of each shift. We provide structural and numerical results for the relationship between the number of aides in an on-call pool, on-call pool rules, staffing costs, and consistency level. We also demonstrate that using part-time aides can actually improve consistency of care if their on-call pool participation rate is sufficiently high. Vincent W. Slaugh, Assistant Professor, Cornell University, Ithaca, NY, 14850, United States, vslaugh@cornell.edu, Alan Scheller-Wolf, Sridhar R Tayur TB24 109-MCC Analytics in Evidence-Based Practice (EBP) Sponsored: Health Applications Sponsored Session

272