| Student First Name |
Student Last Name |
Advisor Name |
School |
Abstract Title |
Email |
Abstract |
| Francisco |
Aldarondo |
Yavuz A. Bozer |
University of Michigan |
Travel Distance Models for AGV-Based Order Picking Systems |
faldaron@umich.edu |
We derive exact analytic expressions for the expected travel distance in AGV-based order picking system operating under two order assignment rules. Under the random assignment rule an order is assigned to any pick station with equal probability. In contrast, the closest picker rule assigns orders to minimize the total travel distance from all containers to the pick station. Closed-form expressions are presented for different pick station configurations. Finally, the impact of the shape and length-to-width ration of the forward area is assessed through numerical experiments. |
| Lucy |
Aragon |
Laila Cure |
Wichita State University |
An Analytic Approach to Incorporate the Six Aims for Quality in the Analysis of Trauma Care Services |
lgaragon@shockers.wichita.edu |
The Institute of Medicine proposed six aims to guide healthcare quality improvement efforts. However, most healthcare improvement programs still evaluate quality along one aim at a time, effectiveness. This research proposes an analytic approach to incorporate all six aims in the evaluation of healthcare quality. A trauma care setting is used to investigate data requirements, develop the methodology and evaluate its implications. |
| Caleb |
Bugg |
Anil Aswani, Deepak Rajan |
University of California, Berkeley |
Logarithmic Sample Bounds For Sample Average Approximation |
caleb_bugg@berkeley.edu |
The Sample Average Approximation (SAA) method is commonly used to approximately solve stochastic optimization problems, and it often works better in practice than existing theoretical bounds suggest for the number of samples needed to ensure the SAA minimum value is close to the true minimum value. In this paper, we derive new theoretical bounds for SAA that, for certain types of constraint sets, are logarithmic in problem dimension, whereas existing bounds are polynomial in dimension. Our approach characterizes the stability of random instances of the optimization problem using stochastic process theory, and then uses this characterization to construct con dence intervals using concentration of measure techniques. Notably, for single stage stochastic optimization problems, we and that the presence of an L1 constraint yields logarithmic bounds on the number of samples needed. This provides theoretical explanation for the success of SAA for capacity- or budget-constrained problems. |
| Aniela |
Garay Sianca |
Sara Nurre |
University of Arkansas |
A Post Disaster Interdependent Infrastructure Restoration Model |
agaraysi@email.uark.edu |
Over the past years, post-disaster operations research models have become necessary due to the increasing number of disasters. We propose a new optimization model that looks to schedule work crews to restore damaged components on a set of interdependent networks while explicitly considering the availability and restoration of the transportation network which enables work crew movement. |
| Gian-Gabriel |
Garcia |
Mariel Lavieri |
University of Michigan |
Developing Insights For Hard-to-Diagnose Concussions Using Machine Learning |
garciagg@umich.edu |
Concussion, the most common traumatic brain injury, is an emerging public health issue with the potential to develop into serious long-term consequences including neurodegenerative disease. Thus, accurate diagnosis is critical. In previous work, we created a methodology to identify which concussions are most difficult to diagnose. Now, we develop and apply a clustering-based framework to identify subgroups which can guide clinical decisions. |
| Liz |
Halter |
Kayse Maass, Todd Huscha, Mustafa Sir, Kalyan Pasupathy |
Washington University in St. Louis |
Using Discrete-Event Simulation to Evaluate Radiology Process Changes Pre-Implementation |
lizhalter@wustl.edu |
In this study, discrete event simulation was used to model the CT practice of Mayo Clinic’s Saint Marys Radiology Department, including six unique CT machines which each handle a different set of exams, scheduled inpatients and outpatients, unscheduled emergency patients with time-dependent arrival rates, and pre- and post-exam bays. Three operational interventions were tested in the model and the outputs were analyzed to inform implementation into practice. |
| Ngoc |
Nguyen |
Ebru Bish, Douglas Bish |
Virginia Tech |
Optimal Pooled Testing Design for Prevalence Estimation under Uncertainty |
ntn@vt.edu |
Accurate prevalence estimation is essential for surveillance of diseases, and is typically conducted via pooled testing. The pool design, i.e., the number and sizes of testing pools, impacts the estimation accuracy. However, determining an optimal pool design requires an initial estimate of the unknown prevalence rate, and can be challenging for emerging diseases. We develop robust optimization models and characterize the properties of optimal pool designs for surveillance. Our case study suggests that estimation accuracy is substantially improved with optimal pool designs. |
| Destenie |
Nock |
Erin Baker |
University of Massachusetts Amherst |
Sustainable Electric Generation Portfolios: A Multi-Criteria Decision Analysis Framework |
destenienock5@gmail.com |
We evaluate the sustainability of electric generation portfolios, using multi-criteria decision analysis applied to the New England power system. The sustainability of generation portfolios with varying levels of offshore wind, natural gas, hydro, and nuclear, are considered under various preference scenarios using a set of sustainability criteria. We find that when the most weight on air pollution and climate change, adding nuclear capacity is a dominant choice. If avoiding nuclear and conserving water is most important, then retiring oil and nuclear while adding high levels of offshore wind rises to the top. |
| Oluwaseun |
Ogunmodede |
Alexandra Newman, Gregory Bogin |
Colorado School of Mines |
Production Scheduling in Underground Mine Operations Incorporation Heat Loads |
oogunmod@mymail.mines.edu |
Mine production scheduling determines when three-dimensional blocks of ore should be extracted. The accumulation of heat in airways where operators are extracting ore is an issue when designing a ventilation system and, often, the production scheduling and ventilation decisions are not made jointly. Rather, heat is ignored. Our model maximizes net present value subject to precedence, and resource capacities. The model improves production schedules that could increase revenue by lowering ventilation costs. |
| Toyya |
Pujol |
E.A. Clayton, J.F McDonald, P.Qiu |
Georgia Institute of Technology |
Leveraging TCGA gene expression data to build predictive models for cancer drug response |
pujol@gatech.edu |
This project uses machine learning techniques to predict cancer drug response based on gene expression data from patients’ primary tumor tissues and clinical trial information. We considered two cancer drug models: Gemcitabine and 5- Fluorouracil. The results show our models predict up to 86% accuracy. Models trained on multiple cancers outperformed models based on a single cancer model. We found genes that were most informative for predicting drug response were well-known cancer signaling pathways; highlighting the potential significance in clinician decision making. |
| Donald |
Richardson |
Amy Cohn |
University of Michigan |
Simulating the Outcome of Make-Ahead Drug Policies at an Outpatient Chemotherapy Infusion Center |
donalric@umich.edu |
We have developed a discrete-event simulation tool to evaluate a variety of policies for selecting which drugs to pre-mix for outpatient chemotherapy patients. By making patient-specific drugs ahead of appointment time, patient waiting time can be reduced. However, this comes at the risk of incurring waste cost if the patient defers treatment after arriving for their appointment and their drug must be discarded. We utilize data from the University of Michigan Rogel Cancer Center to validate our methods. |
| Karmel |
Shehadeh |
Amy Cohn, Marina Epelman |
University of Michigan |
Analysis of Stochastic Mixed-Integer Linear Programming Models for the Outpatient Appointment Scheduling Program |
ksheha@umich.edu |
We present a new stochastic mixed integer programming model for outpatient scheduling with stochastic procedure durations, minimizing a weighted sum of waiting, idle time, and overtime. We provide theoretical and empirical comparisons to other models in the literature, demonstrating where signi cant improvement in performance can be gained with our model. |
| Lauren |
Steimle |
Brian Denton |
University of Michigan |
Multi-model Markov decision processes for robust medical decision making |
steimle@umich.edu |
Markov decision processes (MDPs) have found success in the evaluation and design of treatment protocols for medical decision making. However, the usefulness of these models is only as good as the data used to parameterize them, and multiple competing data sources are common in medicine. We show that our new model Multi-model Markov decision process (MMDP), which incorporates multiple estimates of the rewards and transition probabilities of an MDP, can be used to design treatment protocols that are more robust to errors in the data. |
| Alejandro |
Vigo Camargo |
Yavuz A. Bozer |
University of Michigan |
Managing the Multi-Node Replenishment Logistics in the Food Supply Chain |
avigo@umich.edu |
The main objective of this study is to develop an optimization model to minimize the cost of the replenishment logistics in the outbound portion of a fast food supply chain by minimizing the number of visits and trucks needed, while meeting service level and capacity limitations. Through the study, the advantages and limitations of alternative delivery methods will be explored in order to identify the ideal method that has a significant impact on the cost of the entire supply chain. |
| Bing |
Yao |
Hui Yang |
Pennsylvania State University |
Markov Decision Process for Sequential Optimization of Additive Manufacturing |
bzy111@psu.edu |
Additive manufacturing enables the creation of complex and freeform geometries, but lacks the ability to perform real-time quality control. The present paper introduces a new sequential decision-making framework for optimal AM control. This, in turn, allows corrective actions to take for repairing incipient defects prior to completion of the build. |