2014 Student Poster Presentations

Using Epidemic Modeling for Contamination in Food Supply Chains

Jessye Bemley

North Carolina A&T State University

As terrorist attacks become more prevalent in our society there is a growing concern about the vulnerabilities associated with our nation’s food supply. A deterministic differential equation model is employed to simulate and individual consumer becoming ill from a contaminated product. The results will show the benefit of assessing food supply chain vulnerability in order to provide safety to consumers.


Validation of Logistic Regression Models for Clinical Prediction of Metastatic Disease

Maria Fernanda Correa

Department of Industrial Engineering

St. Mary’s University, San Antonio, TX

Computed Tomography (CT) and Bone Scan (BS) are imaging tests used for early detection of metastatic disease among prostate cancer patients. We performed statistical validation of predictive models for CT and BS outcomes to help physicians decide when to recommend patients for imaging. We applied several methods for internal and external validation including leave-one-out and bootstrapping. We present results from our analysis and discuss plans for implementation of the models in a smart phone application.

Healthcare Asset Replacement Problem under Technological Change and Deterioration

Emmanuel des-Bordes

Department of Industrial and Manufacturing Engineering

Wichita State University

This paper presents a discrete optimization model for keeping or replacing a group of aging assets that operate in parallel under a limited budget. Numerical results and sensitivity analyses are presented to illustrate the optimal replacement strategies for Magnetic Resonance Imaging (MRI) and Extremity - MRI machines (eMRI).

Physical-Statistical Modeling of Cardiovascular Systems for Optimizing Medical Decision Making

Dongping Du

Department of Industrial and Management Systems Engineering

University of South Florida

A novel statistical metamodeling approach is proposed for efficient computer experiments and optimization of Na models to describe glycol-altered Na gating kinetics in the ST3Gal4  heart. Experimental results demonstrated the efficiency and effectiveness of the presented algorithms, especially for large-scale simulation models that are computationally expensive and time consuming. This study improves the understanding of glycosylation-associated pathology of cardiac disease, and potentially contributes to the development of pertinent therapeutic solutions.

Real-Time Ambulance Redeployment with Fair Workload Consideration of EMS Providers

Shakiba Enayati

Department of Operations Research

North Carolina State University

Redeployment refers to a dynamic relocation of available ambulances to compensate for the loss in coverage due to busy vehicles. Undisciplined repositioning can result in a conflict with personnel interests and induce unnecessary fatigue. We propose a real-time approach to find the best policy with respect to relocating idle ambulances so as to maximize the coverage and also not to impose undesirable workload to EMS providers.  Our approach is composed of running two mathematical models in succession that can report the result shortly. Each mathematical formulation is a linear binary model. The first model deals with maximizing coverage while considering provider workloads in the set of constraints. Having the maximum possible coverage resulting from the first model, we minimize total travel distance in the second model to avoid unnecessary moves. Both models avoid long trips for ambulance redeployment. The dispatching rule used is to send the closest available unit. We study and verify our approach's performance by applying simulation over a large real-data set from Mecklenburg County, NC. Then we compare the simulation-optimization result with result of a static policy for different performance measures.

When does turnover impact project outcomes: when you do not know it’s coming?

Hise Gibson

Harvard Business School

In this paper, we estimate the effect of multiple team membership (MTM) on project outcomes. We find that MTM negatively impacts project completion and that unanticipated turnover negatively impacts project outcomes. Although organizations consider a variety of factors for team composition, they rarely consider the project team’s current saturation level.

A framework for modeling the DCIS progression into IBC

Shadi Hassani Goodarzi

North Carolina State University

Ductal Carcinoma In Situ (DCIS) is arguably a direct precursor of Invasive Breast Cancer (IBC).  Approximately 14%–53% of DCIS turn into IBC, after long follow-up periods ranging from 18 to 30 years (Bayraktar 2013). So about 47% 86% of the DCIS cases are over diagnosed and as a result, treatment can only cause harm for these patients. This framework will allow us to study the progression of DCIS into IBC more clearly and as a result, aiding both patients and doctors in decision making.

A Decision Support System On Surgical Treatments For Rotator Cuff Tears

Weihong Guo

Department of Industrial and Operations Engineering

University of Michigan, Ann Arbor

Ineffective physical therapy for rotator cuff tear patients who still need surgery eventually increases the time and cost of treatment and pain for patients. This research developed a decision support system to help physicians make an early surgery decision by effectively analyzing high-dimensional and heterogeneous patients’ data with mixed-type and missing values.

A Lagrangian Search Method for the K-Median Problem

Joshua Q Hale

H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

We propose an algorithm for solving the non-metric version of the K-median problem, which integrates the ease of computing a solution given a set of Lagrangian multipliers using Lagrangian relaxation with the robustness feature of stochastic search. We prove that this method can close the duality gap for any variant of the K-median problem.

Appointment Scheduling with No-Shows and Cancellations

Shannon Harris

Joseph M. Katz Graduate School of Business

University of Pittsburgh

Patient no-shows and cancellations can be disruptive to clinic operations. Scheduling schemas such as overbooking or overtime slot assignments can assist with mitigating these disruptions. We propose a two-phase scheduling model that accounts for the no-show and cancellation rates across the patient population using patient dependent forecasting models. Patients are scheduled accounting for appointment lead time, the patient’s probability of cancellation and of no-show, the patient’s expected waiting time,  and the doctor’s overtime and idle time.

A Markov Decision Process to Identify Optimal Policies for Stopping a Trial of Labor

Karen Hicklin

North Carolina State University

The cesarean section rate in the U.S. was reported to be 32.8% in 2011, rising from 4.5% in 1970. Due to the increased risk of short-term complications associated with cesarean sections, among other things, there is general consensus that current rates are too high. A dynamic programming decision model was created to provide threshold values for when to continue or stop a trial of labor in order to maximize the expected utility of healthy outcomes for the mother and child.

Cognitive Degradation Modeling for Alzheimer’s Disease (AD) via a Collaborative Degradation Modeling Approach

Ying Lin

University of Washington

Cognitive monitoring and screening is used for early detection of AD that can be operationalized by developing a personalized degradation model to predict the cognitive status over time. The major challenge in building personalized model is the sparsity and fragmented nature of the cognitive data of each individual. This poster will present a novel collaborative degradation model to mitigate this problem that can effectively incorporate domain knowledge into degradation modeling and demonstrate its effectiveness using both simulation studies and real-world dataset.

Minimizing Overdiagnosis in Cancer Screening while Considering

Heterogeneity in Patients' Adherence Behaviors

Mahboubeh Madadi

Department of Industrial Engineering

University of Arkansas

Screen-detected cancers that would not have presented clinically in a patient’s lifetime in the absence of screening are considered overdiagnosed cancers.  The objective of this research is to minimize the probability of overdiagnosis while maintaining the lifetime cancer mortality risk at a threshold.  We also incorporated the heterogeneity in patients’ adherence behaviors into our model. We applied our model to mammography, currently the most effective method for breast cancer screening.

Stochastic Auto-Carrier Loading Problem

Benita Mordi

Texas A&M University

This poster presents the tactical planning regarding the number and type of auto-carriers  required under uncertainty in the demand of vehicles to be loaded on each auto-carrier.  The problem involves using actual vehicle dimensions, government regulations on total height of the auto-carriers and maximum weight of the axles, and safety requirements. The problem is modeled using two-stage stochastic integer programming, and the test results using real data confirm the advantages of using the new model.

Modeling the Complex Interaction between Comorbidities: Breast Cancer and Diabetes

Nisha Nataraj

Industrial & Systems Engineering

North Carolina State University

In 2010, over 200,000 women were diagnosed with invasive breast cancer, and 12.6 million women were affected by diabetes in the US, according to the Centers for Disease Control and Prevention. While we know much about these prevalent diseases  individually, little has been studied about the interaction between them. Here, we build a modeling framework that explores this complex relationship so as to assess the prognosis for women who are diagnosed with  diabetes given their breast cancer risk.

An Optimization Framework to Improve Patient Safety in Radiation Therapy Care Delivery

Pegah Pooya

North Carolina State University

We develop an optimization framework to improve the process reliability and patient safety of radiation therapy care delivery process for cancer patients. The use of people-based Safety Barriers (SB) in radiation oncology is a widely recognized method for detecting potential errors before they reach the patient. In this study, In this study, a Dynamic Programming model is developed to optimize the location and elements of Safety Barriers (SB).

The Impact of Choice Modeling on Outcomes in a Colorectal Cancer

Rachel Townsley

North Carolina State University

Simulation modeling is a powerful tool in healthcare policy research and can provide valuable insights and predictions.  However, even perfect models of disease progression and treatment efficacy can be dramatically hindered by assumptions about patient choice and behavior. Here we utilize different choice models to govern patient compliance and screening mode selection in a colorectal cancer screening simulation and their impact on health outcomes.

Explaining Contagion through a Game of Borrow and Lender Countries

Jonathan W. Welburn

Department of Industrial & Systems Engineering

University of Wisconsin - Madison

Economic contagion, the process by which crises spread across countries, generates significant risk to global economic stability. Contagion is often attributed to either a trade channel or a debt channel. We use a multi-agent model of heterogeneous countries to explain how risk can be transferred from country to country. Furthermore, we use this model to explain the possibility for contagion-like effects from a common economic cause.