2017 Student Poster Presentations

INFORMS 2017 Minority Issues Forum 6^th Poster Competition Finalists:
(top row: Gabriel Zayas-Caban, Maria Mayorga, Lewis Ntaimo)
(bottom row:  Lauren Steimle, Gian-Gabriel Garcia, Donald Richardson)

*presenting author listed

Volunteer convergence is one of the biggest challenges for relief center managers. To better understand this phenomenon and find effective management strategies, an agent-based simulation model is developed consisting of donors providing relief items, beneficiaries in need of relief items and random arrivals and departures of volunteers. We investigate volunteer assignment policies that reduce donor beneficiary waiting time under given changes in volunteer capacity, available inventory, and beneficiary’s arrival rates.

This study models the global vaccine market for a set of antigen bundles from different manufacturers. Those are sold to international group-buying entities (GAVI, PAHO, etc.) and/or individual countries organized as markets that may or may not cooperate. Through a sequence of optimization problems, vaccine prices and allocations are compared in different scenarios of cooperation to maximize social welfare and ensure profit for the manufacturers.

Highway traffic is an expensive phenomenon that leads to delayed travel time for drivers. Express lanes are optional toll-based roads that aid in managing congestion. However, determining the optimal price that maximizes profit and utility of Express lanes is difficult. We set out to design a flexible and robust One Lane Express System model.  We determine the optimal price is a function of the operating costs, marginal costs and how much the driver's value the Express Lane.

Non-alcoholic steatohepatitis (NASH) is correlated with obesity, however the temporal relationship from obesity development and NASH related cirrhosis is unclear. We aimed to determine the temporal trend behind NASH related additions to the liver transplant waitlist and the rise in obesity in the US population to make projections about the potential burden of NASH over the next decade.

Healthcare providers often rely on decision making heuristics and universal treatments to address post-diagnosis strategies of chronic conditions such as high blood pressure. This lack of personalized treatment fails to account for the patient’s individual characteristics, which can have adverse effects on certain vulnerable populations among which young adults are included. This research proposes a decision support tool, using operations research techniques to assess treatment scenarios presented through for specific simulation generated individuals based on stage of the disease, nature of the condition, likelihood of related complications, and resource availability.

In this study, we use a simulation model aimed to analyze the quality of different potential layouts by using variable neighborhood search. The main purpose of the simulation is to efficiently replicate the customer behavior, in order to accurately capture the customers flow in the supermarket. The simulation evaluates the layout by determining the visibility of the impulse items by predicting customer movement across the store, and using a shelf allocation model to give the generated layout a score.

This research is part of the University Food Systems project aimed at designing cost effective supply chains to bridge the gap between local farmers and universities in North Carolina. We developed transportation schedules that minimize the total distance traveled by farmers to drop off their produce at cross docking facilities.

We use stochastic programming to determine decision thresholds for medical diagnosis risk models under uncertain data samples from a fixed population. These thresholds maximize sensitivity and specificity under constrained false positive and false negative rates. We show that sample average approximation solutions for these models are consistent estimators. We apply our method to concussion assessment and show that our thresholds dominate single decision thresholds under multi-criteria evaluation.

Chicken meat is a high consumed agricultural product and the main source of protein in the world. To supplement demand during a sanitary crisis, i.e. Avian Influenza outbreak, this study will analyze the two supply chains to determine how fast the demand for chicken products can be fulfilled with rabbit meat.

In this study, activity recognition algorithm was developed to distinguish 10 different stationary and moving activities. The data was collected by an Equivital chest sensor that provided triaxial accelerometer data, as well as electrocardiogram (ECG) data. There was also data collection from an Activpal triaxial accelerometer placed on the right thigh. The study consisted of 30 healthy participants (n=15 female, n=15 male) with ages ranging from 19-51 years. There were 108 accelerometer and 10 ECG features derived from both sensors’ raw data.

Heavy metals present in waste water is a major concern in environmental pollution due to globalization and rapid industrialization. The methods for removal of heavy metals from industrial waste waters are chemical precipitations, conventional adsorption, ion exchange, membrane separation methods and electro-remediation methods. Chemical precipitation is economical but low efficiency. This research will assess the dual objective of efficiency and cost of the removal methods in different industrial scenarios, to determine preferred cost-efficient waste water management strategies.

We are the first that incorporate complementarity demand functions in the multi-commodity capacitated facility location problem. We formulate this problemas a 0-1 mixed-integer quadratic program with equilibrium constraints. In common cases where the demands linearly follow the prices, we reformulate the problem and propose an efficient branch-and-refine algorithm. We show analytically and numerically such incorporation lead to superior decisions. We offer management insights to environmentally friendly facility location analysis under the “Cap and Trade” regulation.

Consider patients being treated for a disease (such as Sepsis), who’s condition changes over time. Resources allocated to a patient influence disease progression and outcomes. Our goal is to allocate limited number of resources to patients, depending on their health states, to maximize outcomes. We formulate an MDP model with resources allocated dynamically, and a Jackson queueing network, where the number of resources is assigned in advance. The performance is then tested against existing practices and heuristics.

We study queues with delayed information. Delays in information are very common in service systems in healthcare and transportation settings.  Moreover, they have been shown to cause Hopf bifurcations where the queue lengths begin to oscillate back and forth.  We develop novel approximations for understanding the size of the amplitude of the oscillations.  Our analysis yields new insight for queues with delayed information and how managers of service systems should provide information to customers.

The research will explore the impact of teen pregnancy on the health and wellness outcomes of infants. Causal effects of teen pregnancy were found by matching teen and adult mothers and performing a pairwise comparison for each outcome. The mothers were matched using demographics, health status, and Medicaid eligibility. Outcomes of the infants were restricted to the first year of birth and included addiction rates, health status, foster care, number of Emergency department and wellness visits. 

In collaboration with the University of Michigan Comprehensive Cancer Center, we have developed a predictive model to determine the probability that a patient will not show for or defer chemotherapy treatment. These predications are valuable in managing clinical operations, such as scheduling staff, patients, or determining which drugs to make-ahead.

Markov decision processes (MDPs) have found success in the design of treatment protocols for chronic diseases. However, the usefulness of these models is often only as good as the data used to parameterize them. We introduce the Multi-model MDP as a way to incorporate multiple models of parameter estimates for the design of robust treatment protocols. We illustrate the utility of this approach using a case study of blood pressure and cholesterol control for preventing heart attacks and strokes.