2015 Informs Annual Meeting

SA32

INFORMS Philadelphia – 2015

3 - Balanced Tree Partitioning with Compact Mathematical Programming Formulations Onur Seref, Virginia Tech, 2060 Pamplin Hall (0235), Blacksburg, VA, 24061, United States of America, seref@vt.edu, J. Paul Brooks In this paper, we study mathematical programming formulations for the balanced subtree partitioning problem, some variations of which are known to be NP- complete. We introduce compact exact mixed integer programming formulations for different variations of the problem. We explore the effect of different sets of constraints on our formulations and provide comparative computational results among our formulations as well as other exact formulations and approximation algorithms.

Jack Homer, Owner, Homer Consulting, 72 Station Hill Rd., Barrytown, NY, 12507, United States of America, jhomer609@gmail.com, Gary Hirsch, Bobby Milstein The ReThink Health Dynamics Model simulates many possible interventions to improve a local health system by lowering costs and deaths and boosting equity and productivity through 2040. It has been carefully calibrated to represent several localities across the U.S. A multi-criteria weighting scheme reflecting local values helps to select a single “best” intervention strategy. The best strategy may also depend upon a locality’s room for improvement with regard to health care and health risks. 2 - Systems Models of Post-traumatic Stress Disorder Navid Ghaffarzadegan, Professor, Virginia Tech, 231 Durham Hall, 1185 Perry Street, Blacksburg, VA, 24060, United States of America, navidg@vt.edu, Richard Larson Little agreement exists about effective methods of screening for PTSD, optimal cutoff values, and even the number of PTSD patients. In a series of modeling efforts, we uncover five vicious cycles that inhibit PTSD treatment, and investigate characteristics of an optimal cutoff value for screening. We then show a population level model of PTSD patients in a military/post-military system. The models represent different levels of complexities of PTSD at individual and societal levels. 3 - Patient Flows in Mental Health Clinics Anne Claire Collin, Graduate Student, MIT, 235 Albany Street, Ashdown House #5052, Cambridge, MA, 02139, United States of America, acollin@mit.edu There are different factors that can hinder access and quality of care for veterans in a mental health clinic, many of which are not related to the treatment itself. The purpose of this research is to analyse the flows of patients in Veterans Affairs’ Mental Health clinics in order to improve access. After a thorough work of understanding patients’ conditions and paths through the different services, simulation is used to find a configuration which improves wait times. 4 - Cost Escalation in Health Systems Dominated by Private Health Insurance Nisa Onsel, Research Assistant, Bogazici University, Bogazici University, Industrial Engineering Dept, Bebek, Istanbul, 34342, Turkey, nisa.guler@boun.edu.tr, Yaman Barlas, Günenç Yöcel Adverse selection and moral hazard are effective in individuals’ health insurance plan choice and healthcare service utilization. The inefficient behaviors have a potential impact on increasing insurance premiums. In the existence of increasing healthcare costs due to advanced health technologies, cost escalation in health systems is a complex problem. A system dynamics model is constructed and analyzed for understanding financial sustainability of health systems dominated by private insurance. Chair: Ebru Bish, Associate Professor, Virginia Tech, Dept of Industrial and Systems Eng, 250 Durham Hall, Blacksburg, VA, 24061-0118, United States of America, ebru@vt.edu 1 - The Management of Mass Casualty Incident Response Behrooz Kamali, PhD Candidate, Virginia Tech, 250 Durham Hall (MC 0118), Blacksburg, VA, 24061, United States of America, kamali@vt.edu, Douglas Bish In this research we seek to systematically investigate triage and transport of casualties in the aftermath of a mass-casualty incident. We study the structure of the optimal policies using novel models that incorporate available resources, quantity, and mix of casualties. Insights gained allow us to derive special cases that can be solved to optimality with simple heuristics. We compare results from our models to that of other models in the literature and the current practices. 2 - A Study on the Spatial Spread and Optimal Control of the 2014- 2015 Ebola Outbreak in West Africa Esra Buyuktahtakin, Assistant Professor, Wichita State University, 1845 N Fairmount, Wichita, KS, 67260, Wichita, United States of America, Esra.Buyuktahtakin@wichita.edu, Eyyub Kibis, Emmanuel Des-bordes We develop an optimization approach to capture the disease dynamics of the deadly Ebola virus. We illustrate our model on a case study from Guinea, Liberia, and Sierra Leone. Numerical results demonstrate the accuracy of our predictions and suggest that the model can be used as a decision-making tool to optimally allocate resources for epidemic prevention and control. SA34 34-Room 411, Marriott Health-care Decision Making Sponsor: Health Applications Sponsored Session

SA32 32-Room 409, Marriott Data Mining in Medical and Brain Informatics

Sponsor: Data Mining Sponsored Session

Chair: Sina Khanmohammadi, SUNY Binghamton, 4400 Vestal Parkway East, Binghamton, NY, 13902, United States of America, skhanmo1@binghamton.edu Co-Chair: Chou-An Chou, Binghamton University, 4400 Vestal Parkway, Vestal, NY, United States of America, cachou@binghamton.edu 1 - Nonlinear Dimensionality Reduction for Analysis of Electroencephalography Records Anton Kocheturov, Research Assistant, Department of Industrial and Systems Engineering, University of Florida, 303 Weil Hall, Gainesville, FL, 32611-6595, United States of America, antrubler@gmail.com We suggest using nonlinear dimensionality reduction technique called the Local Linear Embedding for analysis of EEG records. This approach enabled us to distinguish between different states of the brain in a more efficient way comparing to the existing machine learning techniques since it is faster and doesn’t require training of the algorithm. We also detected evidence for local linearity of the brain in the resting state and introduced a new model of the brain based on it. 2 - The Gap Statistics for Determining the Number of Linear Autoregressive Modes in a Multimode Model Vahid Tarokh, Professor, Harvard University, 33 Oxford Street, MD347, Cambridge, MA, 02138, United States of America, vahid@seas.harvard.edu, Jie Ding, Mohammad Noshad We consider modeling of non-stationary stochastic-processes using a multi-mode linear autoregressive (AR) model. Each process is modeled as a mixture of unknown AR components (modes) with an unknown number of modes. We are interested in online determination of the number of modes, and also in identification of each mode. A new model selection approach based on GAP statistics for this purpose is introduced with applications in modeling and prediction of financial and biological data, etc. 3 - Online Machine Learning Framework for Detecting EEG Abnormalities Sina Khanmohammadi, SUNY Binghamton, 4400 Vestal Parkway East, Binghamton, NY, 13902, United States of America, skhanmo1@binghamton.edu, Chou-An Chou Electroencephalogram (EEG) signals provide insight about human brain function. EEG signal abnormalities can be a sign of mental disorders such as depression, epilepsy, etc. Hence, detecting EEG abnormalities can be beneficial for diagnosis, prognosis, and overall improvement of patient’s quality of life. In this work, we present a new online classification framework to detect EEG signal abnormalities in real time. The proposed framework is validated using public EEG datasets. SA33 33-Room 410, Marriott Health Policy/ Public Health Sponsor: Health Applications Sponsored Session Chair: Ozge Karanfil, PhD Candidate, MIT Sloan School of Management, 100 Main Street, E62-379, Cambridge, MA, 02142, United States of America, karanfil@mit.edu 1 - Multi-Criteria, Multi-Place, Multi-Time: Evaluating System Strategies with the Rethink Health Model

48