Portraits of SURE 2025 from CECS and CEHHS

The Summer Undergraduate Research Experiencelogged another big year in 2025, with faculty and student researchers teaming up for more than 40 research projects from UM-Dearborn's four colleges and the Mardigian Library. This year really showcased the breadth and relevance of all the work going on at UM-Dearborn, with projects focusing on cancer research, human rights, cryptocurrency, information transparency and artificial intelligence. Recently, we talked with two student-faculty pairs from the College of Engineering and Computer Science and College of Education, Health and Human Services about what they learned from their projects this summer. Next week, we'll have student-faculty stories from the College of Business and College of Arts, Sciences and Letters.

Robyn Fletcher and Assistant Professor Yiwan Ye
Project: Unequal healthcare access: Children of single mothers of color before and after the Affordable Care Act

By some measures, Robyn Fletcher and Yiwan Ye were an unconventional match for SURE 2025. Ye is an assistant professor of health and human services and Fletcher is a senior studying cybersecurity. But when Fletcher saw the posting for Ye's SURE project, which was a data-heavy investigation of health disparities among single mothers of color, she actually thought it could be a great fit. Her summer internship was also in the healthcare space. And the topic of Ye's project felt close to home. "My mom is a single African American mother, and my mom had issues with healthcare. I even had a close friend, whose mom is a woman of color, and she was unable to get my friend healthcare access when she was younger," Fletcher says. "So I definitely felt interested in getting to learn more about these disparities."

Ye has actually been working on projects related to this topic for nearly a decade. The core work involves analyzing a large dataset from Princeton University that has been tracking health outcomes for mothers and their children since the 1990s. This particular SURE project focuses on how the Affordable Care Act impacted children of single mothers of color, and Fletcher spent much of her time doing a literature review of the most contemporary research in this space. She says it was definitely a change-up from her usual work in cybersecurity. "I'm in CECS, so we don't do a lot of writing. Going through these academic articles was definitely new for me, but it was a great thing for me to learn. I'm already thinking about how I can apply that to my work in CECS," she says.

Ye and Fletcher's work this summer yielded some interesting preliminary findings. Ye says the effects of single parenthood and a mother's racial background aren't merely additive. Rather, these factors often interact in unexpected ways and create unexpected outcomes. For example, white mothers with partners had the highest rates of insured children. But among single mothers, Hispanic single mothers had children with the highest rates of routine care compared to other racial groups. Moreover, Hispanic single mothers had higher rates of routine care for their children compared to Hispanic mothers with partners. "So this is very, very interesting," Ye says. "There is a phenomenon called the Hispanic Paradox, but generally, this refers to how the health of children of Hispanic migrants tends to be better than their American peers, and this can be explained by a number of factors. So what we're observing with single Hispanic mothers could be a new phenomenon."

This is Ye's first time participating in SURE, and he has plenty of praise for both Fletcher and the program. He says the SURE framework, which provides student researchers with workshops on the fundamentals of research and other professional development subjects, freed him and Fletcher up to focus on their core work. "And to be honest, I think without Robyn's help, the project would just sit there," he says. "She gave me the energy, and I was motivated by her passion for this project. And I think her personal and academic background gave me a lot of confidence that this would go well."

Vlad Nitu and Assistant Professor Srijita Das
Project: Uncertainty estimation in deep neural networks

Senior Vlad Nitu hadn't planned on applying to SURE this summer, but when he heard through the Artificial Intelligence Club that Assistant Professor of Computer and Information Science Srijita Das was looking for a research assistant, he jumped at the chance. As someone who's very interested in AI and planning on pursuing a career in that space, Nitu says the topic was right up his alley. Specifically, Das was working on a project involving active learning - a type of machine learning that's very useful when you don't have a lot of labeled data to train your AI models on. Labeling is essential to most AI models and typically involves a human annotator adding tags to data so the AI can learn from them. For example, in a simple image recognition algorithm, an AI model learns from data that has been labeled by a human with various tags, e.g. cat, dog, person, house. But in many contexts, like computer vision, Das says existing labeled datasets are in short supply, and they're very expensive to create.

Active learning, however, allows an AI model to learn even when you have a relatively small pool of labeled data. Das says the magic comes from an approach called uncertainty estimation. First, you train the model on the labeled data you have. Then, you can use various approaches to uncertainty estimation to take an educated guess about which points in a larger unlabeled dataset would be the most useful to label in order to enhance the model's learning. This way, the human annotator can spend time labeling just the most important features in the unlabeled data set, which saves time and money. The newly labeled datasets can then be fed back into the model to provide further training.

Nitu spent the bulk of the summer evaluating different approaches to uncertainty estimation. And the work progressed so quickly, Das then assigned Nitu a bonus project, working on a system to detect "noisy" labels, which occur when human annotators label things incorrectly. "One of the things I learned from doing SURE last year was it's best to keep things a little open ended, because what you are able to get done sort of depends on the student," Das says. "Vlad was so great. He always did his homework. But he would also make his own observations and we'd figure out where to go next. That's a characteristic you see in PhD students. So I think the effort he put in is a big reason we were able to make really good progress this summer." Nitu says he's probably still more likely to go into the workforce than pursue a PhD after he finishes his undergrad. But as someone who's taken some classes in AI but nothing this in-depth, he says getting in the weeds this summer helped him clarify his next moves. "It's definitely helped me lean into AI a lot more, and after this experience, honestly, I'd say this is probably where I want to take my career."

###

Want to learn more about what students did during SURE 2025? Check out the SURE Showcase on Wednesday, Sept. 10 in the Renick University Center's Kochoff Hall. Story by Lou Blouin

Share

• facebook
• twitter
• linkedin