![Show Menu](styles/mobile-menu.png)
![Page Background](./../common/page-substrates/page0405.png)
INFORMS Nashville – 2016
403
WB24
109-MCC
Optimization in Radiation Therapy
Sponsored: Health Applications
Sponsored Session
Chair: Omid Nohadani, Northwestern University, 2145 Sheridan Road,
Room M233, Evanston, IL, 60208-3119, United States,
nohadani@northwestern.edu1 - Designing Radiation Therapy Criteria With Data-driven
Robust Optimization
S. Nastaran Shojaei, Northwestern University, 2145 Sheridan Road
Room C210, Evanston, IL, 60208, United States,
nastaranshojaei@u.northwestern.edu, Seyed M.R. Iravani,
Omid Nohadani
Radiation therapy planning is an iterative process of optimization and evaluation,
making it both time consuming and not reproducible. Often feasibility is not
attainable for which planners relax some of the criteria. We present a data-driven
robust optimization approach that can provide a new and less sensitive set of
criteria which warrant high quality plans despite some constraint violations
within a realistic range. A large set of clinical data is used to inform the method.
2 - Multicriteria Optimization For Brachytherapy Treatment Planning
Victor Wu, University of Michigan,
vwwu@umich.edu,Marina
Epelman, Michael Herman, Kalyan Pasupathy, Mustafa Sir,
Christopher Duefel
High Dose Rate Brachytherapy (HDR-BT) has become a popular mode of
radiation therapy for its ability to deliver high dose localized to the tumor,
resulting in lower risk of side effects. The goal is to allow the physician to explore
trade-offs via an intuitive GUI with respect to multiple dose-volume criteria (also
known as value-at-risk) among high quality plans. The underlying problem is
non-convex and therefore is not practically solvable. The desire to generate plans
quickly, i.e., within the 30 minutes while the patient is under anesthesia,
motivates solving convex approximations (based on conditional value-at-risk)
instead. Our method is retrospectively tested on various cancer sites.
3 - Automated IMRT Treatment Planning For SBRT Paraspinal Case
Using Prioritized Optimization
Masoud Zarepisheh, Assistant professor/attending, Memorial Sloan
Kettering Cancer Center, New York, NY, United States,
zarepism@mskcc.org, Linda Hong, James Mechalakos,
Margie Hunt, Gikas Mageras, Joseph Deasy
Treatment planning is a patient specific and time consuming task, with plan
quality heavily dependent on planners’ skills. In this study, we are employing
prioritized optimization (PO) to automate the planning process. PO is a step-wise
technique where the highest priority goal (e.g., tumor coverage) is optimized first.
At each iteration step the previous objectives are turned into constraints and a
new goal is optimized. We integrate our optimization package with the
commercial treatment planning system called Eclipse.
4 - Time-dependent Radiation Therapy Optimization
Arkajyoti Roy, Bowling Green State University, 4154 Moser Ln,
Bowling Green, OH, 43551, United States,
aroy@bgsu.edu,
Omid Nohadani
Low oxygen concentration reduces the radio-sensitivity of cells. Re-oxygenation
leads to temporal changes during treatment. However, the re-oxygenation
trajectory is unpredictable, leading to uncertain radio-sensitivity. We develop a
time-dependent uncertainty set that models the evolution of radio-sensitivity. To
reduce over conservativeness at later time-periods, a two-stage robust
optimization approach is proposed that can incorporate such uncertainties. For a
clinical prostate cancer case, the robust method is compared to current clinical
methods.
WB25
110A-MCC
Logistics II
Contributed Session
1 - Coordination Between Shipper And Carrier In City Logistics
Gitae Kim, Hanbat National University, Dept. of Industrial
Management Engineering, School of Engineering, Daejeon, 34158,
Korea, Republic of,
gitaekim@hanbat.ac.kr,Juncheul Park
In a decentralized system in city logistics, two stakeholders such as shipper and
carrier have different their own objectives. Coordination is necessary to obtain the
win-win strategy for two parties. This paper investigates contract models between
a shipper and a carrier to achieve the coordination in city logistics. Quantity
flexibility (number of transportation services) and revenue sharing contract types
are formulated by stochastic programming model using options. From the
experimental results, we find the efficient frontier for two stakeholders.
2 - An Optimization Framework For Simultaneous Space Logistics
Mission Planning And Spacecraft Design
Hao Chen, University of Illinois at Urbana-Champaign,
310 E. Springfield Ave, Champaign, IL, 61820, United States,
hchen132@illinois.edu, Koki Ho
This paper proposes a network modeling and optimization method for human
space exploration campaign-level mission planning. The interplanetary space is
discretized into nodes and the space missions are modeled as generalized multi-
commodity network flows, where payload, propellant, and spacecraft are
considered as separate commodities. This problem results in a mixed-integer
nonlinear programming, and we solve this problem with branch-and-bound and
gradient-based method (e.g., SQP). The proposed framework enables us to
optimize the space mission and its spacecraft design concurrently at a mid-fidelity.
WB26
110B-MCC
Information Systems I
Contributed Session
1 - Designing Referral Policies For Optimal Membership Growth:
A Real World Randomized Experiment In An Exclusive Online
Dating Site
Rodrigo Belo, Assistant Professor, Erasmus University, Mandeville
Building T09-20, Burgemeester Oudlaan 50, Rotterdam, 3062 PA
Rotterdam, Netherlands,
rbelo@rsm.nl, Ting Li
We use data from a real-world randomized experiment in an exclusive online
dating site to study the effect of member-get-member referral policies on
membership growth and online user activity. We find that stricter policies, i.e.,
policies that require members to invite more friends so that they can continue
using the service for free, are more effective at fostering growth in multiple
dimensions, including invitations, online user activity, and paid memberships. We
discuss the mechanisms that may be at play and implications for business.
2 - Agent Based Simulation For Social Support Networks
M. Gisela Bardossy, University of Baltimore,
1420 N. Charles Street, Baltimore, MD, 21201, United States,
mbardossy@ubalt.edu,Stefano Za, Eusebio Scornavacca
Support networks have benefited from digital and networking tools. In an
interconnected world, people depend on each other to achieve their personal and
group goals. We model this inter dependency using agent based simulation and
test various hypothesis regarding rules of engagement, dissipation of information
in the network, discrepancy between reality and beliefs and the updating and
correction of beliefs. This information can inform the design of social platforms as
the implications of choice characteristics are better understood. Some preliminary
results are analyzed and discussed.
3 - Multi-homing Within Platform Ecosystems: The Strategic Role Of
Human Capital
Vijayaraghavan Venkataraman, Georgia Institute of Technology,
800 West Peachtree NW, Atlanta, GA, 30308, United States,
vijayaraghavan.venkataraman@scheller.gatech.edu,Marco Ceccagnoli, Chris Forman
Even though there has been considerable research on platform ecosystems, prior
literature has focused mostly on the platform owners and their strategies. In this
paper, I look at the complementor firms, instead, and try to understand why the
incidence of multi-homing, a strategy in which a complementor firm chooses to
join multiple platforms rather than one, is often quite low. I build a capability
framework based on human capital and test it on micro-level data from the ERP
platform ecosystem. The study has important implications for our understanding
of platform growth and innovation and contributes toward the literature on
platform ecosystems as well as strategic human capital.
4 - The Impact Of Sharing Markets On Product Durability
Maryam Razeghian, Ecole Polytechnique Federale de Lausanne,
ODY 4.16, Station 5, Lausanne, 1015, Switzerland,
maryam.razeghian@epfl.ch, Thomas A. Weber
This paper studies the effects of sharing markets on the prices for new products
and on product design in terms of durability. In a dynamic economy with
overlapping generations, consumers take strategic purchasing decisions,
anticipated by a durable-goods monopolist. Without sharing, the optimal
durability increases in the production cost. In the presence of sharing, the firm
prefers to limit durability for low-cost products, effectively disabling a secondary
sharing market. However, all else equal, a peer-to-peer economy never decreases
the incentives to provide durability.
WB26