Come out and show your love for the OR and ISE Ph.D. students, Erik Rosenstrom (OR), Srinivasan Balan (ISE), Annisa Marlin Masbar Rus (ISE) and Margaret Tobey (OR). They are practicing their talks for the upcoming INFORMS Annual Conference.
OR PRO TIP: All OR 601/801 students must attend in person. Others are welcome to join in person or can join by Zoom.
https://ncsu.zoom.us/j/93958805097?pwd=azltMmloZlp6dzRLWjBtOUtUdU1pdz09
Meeting ID: 939 5880 5097
Passcode: 547206
Erik Rosenstrom | Ph.D. Student, Operations Research
Catching Diabetic Retinopathy Early with Machine Learning
Diabetic retinopathy (DR) is a complication of diabetes that can become vision-threatening (VTDR) and cause blindness. In the US, an estimated 899,000 diabetic adults have VTDR despite it being preventable with timely treatment. VTDR is difficult to catch due to the slow progression and dependence on patients’ care-seeking behavior. Here we address these challenges by leveraging 20+ years of electronic health records to construct and extend ensemble classifiers to (i) identify patients that will develop VTDR within the next year and (ii) identify those that will develop DR in the next year. We can achieve high recall (>75%) for both classification tasks. This classifier can personalize care coordination to improve utilization and timing without additional patient actions.
Srinivasan Balan | Ph.D. Student, Industrial and Systems Engineering
Experimental Evaluation of Chance-Constrained Models for a Capacitated Stochastic Production Inventory System
We consider the problem of planning releases to a capacitated production-inventory system governed by queuing behavior and stochastic demand. We use a non-linear clearing function to represent the queuing behavior of the production system and a shortfall-based chance-constrained (CC) model to obtain scalable approximate solutions. We use a dynamic adaptive decision rule to implement the results of the CC models and evaluate their performance using simulation.
Annisa Marlin Masbar Rus | Ph.D. Student, Industrial and Systems Engineering
A Comparative Communication Analysis of Two U.S. Hospital Systems’ COVID-19 Communications Using Topic Modeling
Although understanding initial internal communication to manage the COVID-19 pandemic within healthcare systems is important for effective operational strategies, little is known regarding the themes within hospital-level communications. This study used structural topic modeling (STM) to analyze the internal communications of two healthcare systems from 2/2020 to 5/2021. The analysis also captured the influence of the hospital systems on communication content. COVID-19 communication topics were characterized using STM and their cross-healthcare systems variation to be compared. Variation was found in the main topics discussed across the healthcare systems in responding to the pandemic throughout the specified timeline.
Margaret Tobey | Ph.D. Student, Operations Research
Interpretable Models for the Automated Detection of Human Trafficking in Illicit Massage Businesses
Illicit massage businesses (IMBs) profit illegally from the labor and sexual exploitation of victim workers. To detect human trafficking in this area, we combine data from multiple internet sources and train interpretable prediction models to identify human trafficking risk factors and assess each business’s risk level.
