DATA SCIENCE UNIT
Putting clinical and population data to work on cardiovascular questions, many of them specific to Arizona.
The Sarver Heart Center Data Science Unit helps cardiovascular researchers answer questions that depend on data: health-system records, long-running population studies and, increasingly, information about where people live.
We turn clinical records into research datasets, analyze how risk factors and disease change over time, and are beginning to connect patients' health to place: heat, distance to care, neighborhood resources. The unit is led by Alane Blythe C. Dy, PhD, MSPH.
Our largest current project is the Arizona Heart Study (AZ-Heart), which asks why heart disease burden varies so widely across the state. More about AZ-Heart.
What we do
Turn clinical records into research data
Routine care produces diagnoses, blood pressure readings, lab results and prescriptions for patients across Arizona. Before any of it can answer a research question, someone has to work through regulatory approval, pull the right records, clean and harmonize them, and manage them securely. That is our work. Phase I of the Arizona Heart Study is being built this way.
Follow health over time
Many of the risk factors cardiologists know today were found by measuring the same people again and again, for years. We analyze that kind of repeated data: how blood pressure, weight, smoking and lung function change through adulthood, and how those paths relate to premature death. Dr. Dy's work with TESAOD, a Tucson population study that began in 1972, is one example.
Connect health to place
An address carries a lot of information: how hot summers get, how far it is to the nearest clinic, how much shade and green space are nearby. We are converting residential locations into map locations (geocoding) so they can be analyzed alongside environmental and neighborhood data. The goal is to learn which features of place are associated with heart disease and its risk factors.
Why follow populations?
Established population-based cardiovascular cohorts have followed large groups of people over time, often decades, to see which personal and environmental factors come before heart disease.
That is how population science works alongside the lab: studies of people show what goes with disease, and laboratory research works out why. Findings from these cohorts have shaped how we prevent, diagnose, and treat heart disease.
Those landmark cohorts were built elsewhere. Tucson has its own long-running population studies, but they were designed to study airway disease, not heart disease. Arizona's heat, distances, geography and population raise questions that need Arizona data.
Flagship initiative: the Arizona Heart Study
The Data Science Unit leads the Arizona Heart Study (AZ-Heart), launched in November 2025. It is the unit's largest project, and it uses all three kinds of work described above.
The question: why does heart disease burden vary across Arizona?
Heart disease is the leading cause of death in Arizona, and the burden is not spread evenly. In the Arizona Department of Health Services' 2022 Heart Disease Burden Report, cardiovascular death rates vary widely from county to county.
Some of that gap reflects who lives in each county. AZ-Heart researchers are examining whether features of place may also help explain it:
- Extreme heat. In Maricopa County, heart and lung disease contributed to 47% of heat-related deaths in 2025.
- Distance to care. Rural Arizonans travel about 20 miles to reach a health care provider; urban residents travel about 3.
- Neighborhoods and green space. Trees and vegetation are unevenly distributed across Arizona's urban neighborhoods.
- A diverse, aging population. About 19% of Arizonans are 65 or older, a share expected to reach 22% by 2035. Communities across the state differ in their backgrounds and in the conditions they live in.
These are factors that may be associated with heart disease burden. AZ-Heart is designed to measure them, not assume them.
What we have now: Phase I (active)
Phase I is a retrospective study: it uses existing health records. The unit is analyzing Arizona-based clinical data to describe heart disease and its risk factors across the state.
It also helps the unit build what a larger study will need: regulatory processes, a data pipeline, and systems for data management and statistical oversight.
Where it leads
- Describe the cohort (underway). Who is in it, and which heart conditions and risk factors occur together, by age and group.
- Link health to place (underway). Geocoding addresses, then analyzing the records alongside environmental and neighborhood measures.
- Phase II: a prospective cohort (planned). A future study that would follow a group of Arizonans over time, with repeated clinical assessments, biological samples and possibly digital or wearable health data. What Phase I shows will shape its design.
The long-term aim is an Arizona cohort that can show how cardiovascular risk and disease develop in the Southwestern U.S. environment.
Read the November 2025 announcement of the Arizona Heart Study
Questions behind the data
Population science starts with specific questions. These are some of the questions shaping the unit's current work.
Why do cardiovascular death rates differ so much across Arizona counties?
Age, access to care and environment are all possible contributors. Separating them takes health records linked to place, which is what Phase I of AZ-Heart is building.
Does living 20 miles from a doctor instead of 3 show up in heart health?
Once addresses are geocoded, researchers can ask whether distance to care is associated with differences in heart disease and its risk factors.
How does extreme heat relate to heart disease?
Heat-related deaths are the most visible cost. A cohort linked to place lets researchers ask a broader question: are heart conditions and risk factors more common where exposure to extreme heat is greater?
Which conditions cluster together, and in whom?
Heart failure rarely arrives alone. Heart conditions and risk factors such as high blood pressure, diabetes and kidney disease occur together, and early Phase I work shows how those patterns differ by age, race and ethnicity.
What would following people over time show that records cannot?
Health records are collected for care, not research, and they exist only when someone sees a clinician. A study that measures people on a regular schedule could show how risk builds before disease is diagnosed. That is the purpose of the planned Phase II.
Work with the Data Science Unit
The unit collaborates with Sarver Heart Center investigators and research partners on questions that depend on clinical, population or geographic data.
If you have a question like that, or data you think could answer one, contact Dr. Dy. Conversations are most useful early, while a study's design can still change.
Our team
Alane Blythe C. Dy, PhD, MSPH
Director, Data Science Unit
Lead investigator, Arizona Heart Study
Drawing on her experience in both laboratory and population science, Dr. Dy works where mechanistic discovery meets patient outcomes: testing molecular findings in clinical or population cohorts and turning patterns seen in people into new biological questions. At Sarver Heart Center, she is building a shared data infrastructure that connects human data with experimental research.
Students
The unit mentors students and early-career researchers in clinical and population data science.
Sumaiya Kaiser, MS
Medical Student, Class of 2029
Minh Le, BS
Post-baccalaureate Researcher