Each year students in the University of Washington Department of Biostatistics Master of Science Capstone program apply their knowledge and skills to help address a wide range of pressing public health challenges.
In 2026, teams collaborated with UW research centers to evaluate trends in treatments for patients with breast cancer and appendicitis. A third project explored whether cell shape is related to cellular function, and whether this relationship can be detected computationally.
“The projects asked really important questions about cancer treatment, clinical care for appendicitis, and basic cell biology,” said Kathleen Kerr, MS Capstone program director and professor of biostatistics. “The students really rose to the challenge. I think they learned a lot and contributed a lot, which is exactly the goal of the Capstone project.”
2026 Projects
Survival Impact of Targeted Vaccine in Inflammatory Breast Cancer
- Sponsor: UW Cancer Vaccine Institute
- Team: Enya Liu, Xiao Shi, Luyi Tong, and Jieyi Xu
Inflammatory breast cancer (IBC) is a rare but highly aggressive subtype of breast cancer with poor survival outcomes. In collaboration with UW Cancer Vaccine Institute, this project evaluates whether a HER2-directed therapeutic vaccine improves survival among patients with HER2-positive IBC. Because the vaccine cohort does not include an internal control group, the team constructed an external comparison cohort from the Surveillance, Epidemiology, and End Results (SEER) registry. Using propensity score matching and survival analysis methods, the team compared overall survival between vaccinated patients and matched controls. Findings suggest substantially improved survival among vaccinated patients, and demonstrate how population-based registry data can support the evaluation of emerging cancer therapies.
Non-Operative Management of Appendicitis — Trends in Selection and Outcomes
- Sponsor: Griffin I. Allen, MD, Mbe, UW Surgical Outcomes Research Center
- Team: Breanna Brown, John Chen, Sam Herold, Anke Li
The CODA trial helped establish antibiotics as a viable alternative to appendectomy for selected patients with uncomplicated appendicitis. Building on this work, the team examined how non-operative management (NOM), or antibiotics-first treatment, has been used in real-world practice across multiple U.S. sites. Using retrospective chart-reviewed data from the TRIAD study, the team evaluated changes in NOM use after implementation of a decision support tool called AppyOrNot. This is a free tool available to patients that explains what appendicitis is, different ways to treat it, and offers guidance based on personal criteria.
The team studied factors related to initial treatment selection, and assessed one-year outcomes among patients who initially chose NOM. Their analysis found that NOM use increased over time, with a larger increase at non-CODA sites after AppyOrNot implementation. The team also found that most patients who initially chose NOM did not undergo subsequent appendectomy within one year, and appendicolith (a calcified, hardened deposit of fecal matter (fecalith) located in the appendix) appeared to remain an important consideration in both treatment choice and downstream outcomes.
CODA trial:https://www.nejm.org/doi/full/10.1056/NEJMoa2014320
TRIAD study: https://becertain.org/about-triad
Exploring Morphology – Gene Expression Associations at the Molecular Level
- Sponsors: Katie Prater, assistant professor, Neurology, UW (scientific sponsor), Kevin Lin, assistant professor, Biostatistics, UW (statistical sponsor)
- Team: Jinqiu Du
The project explored whether cell shape is related to cellular function, and whether this relationship can be detected computationally. Using a publicly available dataset where both cell morphology (shape) and gene expression were measured from the same neurons in the mouse brain, work included transforming the complex morphology data into a numerical representation that captures differences in cell shape, then using a statistical method to identify genes whose expression patterns align with variations in morphology. To interpret these genes, a pathway analysis was applied that checked whether the selected genes corresponded to known biological processes. In addition, a machine learning model that integrates morphology and gene expression into a shared representation was used. Findings suggested that structural features of cells can provide insights into their molecular state, and that computational models can help bridge imaging data and genomic data.
More about MS Capstone project sponsorship
