2016 INFORMS Annual Meeting Program

340 / 561

INFORMS Nashville – 2016
340
2 - Connected Zero Forcing Of A Graph
Boris Brimkov, Rice University,
Zero forcing is a dynamic graph coloring process whereby a colored vertex with a
single uncolored neighbor forces that neighbor to be colored. This talk introduces
the connected forcing process - a restriction of zero forcing in which the initially
colored set of vertices induces a connected subgraph. The connected forcing
number - the cardinality of the smallest initially colored vertex set which forces
the entire graph to be colored - can be used to bound various linear algebraic and
graph parameters, as well as to model the spread of diseases and information in
social networks. Other properties of the connected forcing number are discussed,
and closed formulas are given for several families of graphs.
3 - An Integer Programming Approach To Finding Minimum
Zero-forcing Sets
Caleb Fast, Rice University, , Illya V Hicks
In this talk, we introduce an integer programming approach for finding minimum
zero-forcing sets. Zero-forcing is a graph coloring process that, in addition to
applications in power networks and quantum computing, is used to bound the
minimum rank of matrices characterized by the given graph. Straightforward
integer programming formulations of this problem do not perform well; however,
we incorporate new bounds on the zero-forcing iteration index and zero-forcing
number to improve the performance of the integer program.
4 - Mathematical Programming Approaches To Influence
Maximization Problems On Social Networks
Rui Zhang, University of Colorado - Boulder, Boulder, CO,
United States, Subramanian Raghavan
We study influence maximization problems on social networks from an integer
programming perspective. In this talk, we focus on the weighted target set
selection problem. Motivated by the desire for exact approaches, a tight and
compact extended formulation is presented on trees. A complete description of its
polytope is given as well. Furthermore, based on the extended formulation, a
branch-and-cut approach is proposed for general networks. Computational results
based on large scale graphs are discussed. Lastly, we present our results for
different variants and generalizations of influence maximization problems.
TD20
106C-MCC
Optimality Conditions for Inventory Control
Invited: Tutorial
Invited Session
Chair: Eugene A Feinberg, Stoney Brook University, Department of
Applied Mathematics, Stoney Brook, NY, 11794, United States,
1 - Optimality Conditions For Inventory Control
Eugene A Feinberg, Stoney Brook University, Department of
Applied Mathematics & Statistics, Stoney Brook, NY, 11794,
United States,
This tutorial describes recently developed general optimality conditions for
Markov Decision Processes that have significant applications to inventory control.
In particular, these conditions imply the validity of optimality equations and
inequalities. They also imply the convergence of value iteration algorithms. For
total discounted-cost problems only two mild conditions on the continuity of
transition probabilities and lower semi-continuity of one-step costs are needed.
For average-cost problems, a single additional assumption on the finiteness of
relative values is required. The general results are applied to periodic-review
inventory control problems with discounted and average-cost criteria without any
assumptions on demand distributions. The case of partially observable states is
also discussed.
TD21
107A-MCC
Data Analytics for Healthcare
Sponsored: Health Applications
Sponsored Session
Chair: Qingpeng Zhang, City University of Hong Kong,
83 Tat Chee Ave, Kowloon Tong, TX, 00000, Hong Kong,
1 - Bayesian Data Analytics For Individualized Health Care Demand
Modeling Of Aging Population
Xuxue Sun, University of South Florida, 4202 E. Fowler Ave.
ENB118, Tampa, FL, 33620, United States,
Paul Cirino, Hongdao Meng, Nan Kong,
Mingyang Li
With high risk of having health problems, elderly people are mainly users of
health care services. Successfully modeling of their demands will facilitate
decision makings in health care management. Existing approaches mainly utilized
aggregate-level data from single type of health care facility and studied the
observed factors’ influence. In this study, a Bayesian data analytics approach
accounting for competing risk of different facilities is proposed to characterize
individualized health care demand and to jointly quantify both unobserved
individual heterogeneity and observed factors’ influence. A real case study is
further provided to demonstrate the effectiveness of proposed method.
2 - Harnessing The Power Of Twitter With Offline Contact Networks
For Probabilistic Flu Forecasting
Kusha Nezafati, University of Texas, Dallas, TX, United States,
, Qingpeng Zhang, Yulia Gel,
Leticia Ramirez-Ramirez
The prompt detection and forecasting of infectious diseases are critical in the
defense against these diseases. Despite many promising approaches, the lack of
observations for near real-time forecasting is still the key challenge for operational
disease prediction and control. In contrast, online social media has a great
potential for real-time epidemiological forecasting and could revolutionize
modern biosurveillance capabilities. We investigate utility of Twitter to serve as a
proxy for unavailable data on flu occurrence and propose a predictive platform
for disease dynamics by accounting for heterogeneous social network interactions,
space-time, and socio-demographic information. .
3 - The Diffusion Of User Behavior via Social Network In Online
Health Communities
Xi Wang, University of Iowa, Kang Zhao,
Gautam Pant
As a major source of social support for people with health problems, Online
Health Communities (OHCs) have attracted a great number of members. Prior
research has examined that users involved in online community motivated either
by community-interest or by self-interest. Using text mining and unsupervised
machine learning techniques, we revealed that users of a popular breast cancer
OHC acting different roles corresponding to their motives. We also found user
role can diffuse via social ties. Our research has implications to for OHC operators
to track users’ behaviors in order to manage an OHC.
TD22
107B-MCC
Decision Making in Healthcare Supply Chain
Sponsored: Health Applications
Sponsored Session
Chair: Mili Mehrotra, University Of Minnesota, 321 19th Ave S,
Minneapolis, MN, 55455, United States,
1 - Gatekeeper Or Roadblock: Optimizing Evidence Generation
Andgatekeeper Or Roadblock: Optimizing Evidence Generation
And Access To New Drugs
Liang Xu, Pennsylvania State University, 419A Business Building,
Penn State University, State College, PA, 16801, United States,
, Hui Zhao
In 1992, the accelerated approval pathway (AP) is was instituted to speed up the
development of new drugs but failed to be effective due to sponsors’ lack of
incentives to complete post-market study. We propose and analyze three
mechanisms, i.e., extra market exclusivity, pay for evidence, and augmented user
fee to incentivize post-market study. Our results provide insights for policy
makers on granting accelerated approval with the consideration of post-market
study.
2 - Hospital Quality, Medical Charge Variation, And Patient Care
Efficiency: Implications For Bundled Payment Reform Models
Seokjun Youn, Texas A&M University, College Station, TX, United
States, , Gregory R Heim, Subodha Kumar,
Chelliah Sriskandarajah
We examine how unwarranted variation in hospital medical charges relates to
patient-centric goals. From a policy maker’s viewpoint, the results imply that
managerial incentives based on process quality (rather than outcome quality)
may be more effective for changing operational behaviors that lead to lower
variation and higher efficiency. We investigate these implications for bundled
payment programs. Empirical results suggest that the current bundled payment
provider selection mechanism does not consider the degree of unwarranted
variation in charges, which we claim to be the improvement opportunity for each
participating provider.
TD20