Xue said her fascination with statistics began almost by accident. As an undergraduate in China, she enrolled in a program that combined finance and statistics, intending to pursue the finance track. But her early exposure to statistical methods quickly won her over.
“It’s a beautiful combination of theory and application. It’s mathematically deep so you feel satisfied, but also you’re going to have interesting applications. You can see its impact and how that method is going to be used in real life,” she said.
Xue’s statistical background is in non-parametric methods, which allow data to reveal its own patterns without assuming a specific model form. She began her work developing methods for dimension reduction and model selection, helping make high-dimensional “big data” more manageable for analysis.
“Non-parametric is the backbone, it’s my hammer. I use it to tackle different kinds of problems,” she said.
Over time, she expanded her research to complex data types, including brain imaging networks and functional data, such as physical activity tracked by wearable devices. These devices don’t take exact measurements, but instead rough estimates. Statisticians like Xue are needed to figure out how to deal with errors to make sure the data is as robust as possible.
Building on this expertise, Xue is contributing to an NIH-funded project aimed at improving the accuracy of self reported diet and activity data in obesity and type 2 diabetes research. The grant, awarded by the National Institutes of Diabetes and Digestive and Kidney Diseases, brings together an interdisciplinary team led by Carmen Tekwe, associate professor at Indiana University.
The researchers are tackling a critical challenge: how to correct the errors that occur when using wearable devices such as fitness trackers and self-reported dietary questionnaires. These tools are widely used in large-scale studies but are prone to recall bias, systematic error and complex correlations that can distort the statistical models used to evaluate health outcomes.
By designing new methods to account for these problems, the team aims to provide a more reliable picture of how diet and physical activity influence obesity and type 2 diabetes. Their work has the potential to improve public health recommendations, strengthen the evidence used for chronic disease prevention and ultimately lead to more effective interventions tailored to diverse populations.
"Statistics a beautiful combination of theory and application"
Her research prowess is matched by her commitment to collaboration. Xue believes statistics is strongest when applied to real-world challenges, and many of her projects are problem-driven, sparked by the needs of collaborators in medicine, biology and agriculture. One applied paper, honored with a Best Paper Award from the Review of Regional Studies, examined how administrative costs vary across U.S. counties depending on governance structure, population, urban versus rural status and natural amenities. Using advanced statistical methods, the study revealed that factors like wages, population size and health outcomes affect operational expenditures differently across counties, underscoring the importance of local context in policy and resource allocation. This research grew from a partnership with faculty in the College of Agricultural Sciences.
Looking ahead, Xue is preparing her department to engage with the rise of artificial intelligence. She sees two opportunities: using AI tools like deep learning to tackle longstanding statistical problems, and infusing AI development with rigorous statistical reasoning to improve uncertainty quantification and inference.
“We are ready to adapt and adjust. There are many ways that statistics can contribute to AI,” she said.
Outside of work, Xue enjoys traveling and cheering at her son’s soccer games, moments of joy that balance her busy academic life.
