As healthcare rapidly digitizes, the ability to translate vast, complex medical datasets into actionable life-saving insights has become essential. Health data science bridges this critical gap — transforming big data into targeted treatments and improved public health strategies.
To meet this growing demand, Brown University launched the online Master of Science in Biostatistics program in 2024. This summer, its inaugural graduating class completed their capstone projects, earning degrees from the School of Public Health.
The online master’s program offers a rigorous 20-month curriculum tailored for working professionals with a concentration in Health Data Science. As a culmination of the program, the capstone challenges students to apply their coursework to a real-world problem of their choosing.
The biostatistics capstone project
For their capstone project, students must exhibit independence in selecting their research question and statistical methodology, write a long-form report and executive memo, generate compelling data visualizations and present the results. All of this is carried out in a team-like environment, where students work together, gathering peer feedback while work shopping their projects.
“The diversity of student capstones was really exciting to see!” said Peter Lipman, associate professor of the practice of biostatistics and capstone instructor. “Some students were exceptionally ambitious, obtaining complex health datasets from their current employment or field of interest. Many chose to perform meta-analyses, which means they first performed extensive literature reviews, and then combined their results using mixed effects models to better quantify effect sizes and possible heterogeneity.”
While producing journal-ready research was not a strict requirement, Lipman noted that several students chose to pursue publication: “I am thrilled that the capstone provided them a platform to achieve their goals.”
Below are highlights from the first cohort’s “Problem-driven Biostatistics and Capstone Project” for the online Master of Science in Biostatistics, Health Data Science program.
Veronica Leary: Monitor Displacement in Disaster Health Surveillance
Leary, a dual Brown graduate who completed her online MPH in 2025, investigated how Hurricane Maria impacted population displacement in Puerto Rico.
Her analysis revealed that population flight was the single clearest indicator of storm damage: in the hardest-hit municipalities, departures among Medicare beneficiaries surged by nearly 11% for every one-standard-deviation increase in storm severity. Interestingly, traditional metrics like hospital visits and overall mortality rates showed no clear link to storm severity. However, this does not reflect an absence of health crises; rather, when residents flee affected areas, they disappear from local health records, artificially stabilizing local healthcare usage numbers.
To address this surveillance gap, Leary recommends a 90-day pilot program to track residential mobility alongside hospital usage and mortality data. The system would actively flag missing or incomplete records to ensure population displacement isn't mistaken for a lack of health needs.
Tingwei Adeck: Sensitivity Analysis of the Relationship Between COPD and Psychological Distress
Agriculture and its supporting industries contributed nearly $1.5 trillion to the U.S. Gross Domestic Product in 2023. However, as a leading global source of air pollution, the industry exposes agricultural workers to significant respiratory risks—most notably Chronic Obstructive Pulmonary Disease (COPD).
Adeck’s study examined these risks and found that psychological distress, smoking habits and age were critical factors driving the odds of developing COPD in farmers.
As a result, Adeck recommends policy initiatives that fund research into the nature of psychological distress, a meta-analysis of multiple studies linked to COPD and targeted programs to curb tobacco use and improve quality-of-life among aging U.S. farmers.
Stephanie Anderson: Autoimmune Vulnerability Index (AVI): findings and recommended next steps
Anderson evaluated the Autoimmune Vulnerability Index (AVI), a predictive tool designed to forecast where chronic and autoimmune diseases will rise, rather than simply identifying existing high-risk areas. Her research demonstrated that the AVI outperforms three federal risk-assessment tools currently in use.
To translate these findings into practice, Anderson proposes a 12-month pilot program across 5 to 10 high-risk counties identified by the AVI.
The pilot would test whether deploying resources—such as targeted health programs or new clinics—yields measurable improvements in health outcomes. Additionally, she recommends continuously refining the index by incorporating new economic, environmental and health metrics while recalibrating category weights annually.
Kaity Trinidad: Using Population Profiling to Strengthen Behavioral Health Surveillance
Standard behavioral health surveillance often relies on aggregate statistical averages, which can obscure critical nuances between patient groups—for instance, two states might have similar overall treatment numbers but serve very different types of patients with different needs.
Using Latent Class Analysis on existing substance-use treatment data, Trinidad identified six distinct patient treatment profiles without requiring new data collection. She recommends that agencies like the Substance Abuse and Mental Health Services Administration (SAMHSA) adopt population profiling to complement current surveillance efforts, giving public health leaders a clearer, actionable picture of the populations they serve
Brandon Yee: Recommendation for a Two-Option Blood-Based Alzheimer’s Disease Testing Strategy
Yee evaluated whether blood-based biomarkers could streamline early Alzheimer’s disease screening, reducing reliance on invasive spinal fluid tests or costly PET scans. His analysis focused on two biomarkers—plasma p-tau217 and the p-tau217/Aβ42 ratio—both demonstrating strong diagnostic accuracy (88–91% sensitivity and specificity).
Yee recommends offering clinicians the flexibility to use either blood test as an initial screening tool based on patient context. This strategy simplifies early detection while reserving secondary, expensive imaging for complex or ambiguous cases.