Dynamic data duo advances health research

The latest health research promises futuristic treatments, from cancer vaccines to bioengineered organs for transplants to medical nanobots. While these technologies may one day be routine, they are still decades away from being available to the average American.

But researchers also have an opportunity to make much faster advances in care by analyzing the wealth of clinical data gathered at hospitals and doctors' offices, bringing actionable improvements in a matter of months instead of years.

Health care data represents 30% of all information generated worldwide, but up to 97% of data from hospitals alone goes unused, according to the World Economic Forum. Why? For academic researchers, this sensitive personal information isn’t in a format that’s ready for their work. Doctors, on the other hand, don’t typically learn the research or data science skills needed to effectively use data to guide their practices.

Arizona State University is working to break down these barriers in a collaboration with Valleywise Health, a Phoenix-area health care system that provides comprehensive services, particularly for underserved communities. Together, ASU and Valleywise Health are analyzing a plethora of health data to better serve Arizonans.

“Our clinicians have a strong background in patient care. They may be very interested in research. However, they may not have strong training in clinical research analysis,” says Lora Nordstrom, director of clinical research at Valleywise Health. “Patients benefit from improved care when evidence-based medicine is practiced. This is true for the development of drugs and devices, but even more immediately for physician-initiated studies that answer questions clinicians see in their daily practice.”

This effort was made possible by a nearly $1 million grant from the Flinn Foundation, which supports efforts to improve quality of life for all Arizonans through educational enhancements, bioscience advances and other initiatives.

This collaboration is one piece of a complex puzzle to meet health care demands and improve access to services in Arizona.

“ASU is set up in a beautiful way to work with multiple clinical partners across Arizona,” says Kathryn Claypool, health research data manager at ASU’s Research Technology Office who oversaw the efforts to get Valleywise Health data ready for research. “Being the central academic institution can help bridge some of the gaps so health care businesses can cooperate and share data. It's our main aim that no matter what we do here, we serve our community.”

Electronic health record data is usually not generated with research in mind. One obstacle is the sheer quantity and range of data from years of patient interactions. The information can also be incomplete and inconsistent. Additionally, data systems evolve over time and are often organized in ways not easily understood by those working outside of the health care system.

For example, data fields where a researcher might be expecting a number could instead contain a word, letter, symbol or nothing due to changes in the system over time.

“There’s no world where you have perfect data from the start,” says Sean Dudley, associate vice president in the ASU Knowledge Enterprise and the leader of the Research Technology Office. “Our research data management team handles data acquisition, preparation, management and safety so we can bring data of value together and investigators can hit the ground running. We reduce friction and speed up the path to translation so that patient health outcomes can improve in the near future.”

ASU research data managers worked with Valleywise Health to understand what data is being collected, where that data can be found in the electronic medical record and how that information is represented. Then the team began organizing the data into unified data sets to meet specific research goals.

Security is also crucial when working with private health information. Building on ASU’s work to become a trusted third party for Arizona health data, the two institutions collaborated to make sure the consolidated data protected patients’ privacy. The shared data is securely transferred and stored in the Knowledge Enterprise Secure Cloud and is only accessible by authorized users. Valleywise Health and ASU also ensured the research adheres to ethical and regulatory standards in addition to consent and privacy laws.

Once the data was ready for research, new challenges emerged in asking the right questions to uncover valuable insights. For example, if researchers are interested in studying heat-related illnesses in Arizona, that information may not be noted in a way that can be captured in the research dataset — for example, in a physician’s note rather than in a field in a form. Instead, the researchers may need to make inferences about which patients were seen for heat-related illness based on the available data.

“A lot of times we realize what we need actually does not exist,” says Li Liu, an associate professor of biomedical informatics in the ASU College of Health Solutions who also works in the ASU Biodesign Center for Personalized Diagnostics. “There’s a gap between the clinical question and the health data.”

Liu participated in the project’s “doctoral twinning” opportunity, which pairs ASU faculty and student researchers with Valleywise Health doctors and medical students to conduct data-driven research.

“This opportunity gets students connected directly to field scientists to apply what they learned in the classroom to something meaningful,” Liu says.

Tatiana Patton was just starting her doctorate work in biomedical informatics when the doctoral twinning projects began. She was eager to apply her data science skills to the issue of rapid weight gain in newborns, which can lead to childhood obesity and diabetes.

Liu and Patton worked with Dr. Dean Coonrod, the chair of obstetrics and gynecology at Valleywise Health and a professor at Creighton University, and others to analyze data from 14,667 birth records and more than 2,500 other records from well-baby checkups and maternal exams.

Liu guided Patton on analyzing the data, and they worked with Coonrod and Creighton medical students to interpret the results. The team looked at trends in demographic backgrounds, a history of diabetes and other pregnancy complications, and early feeding practices.

Surprisingly, the team found that rapid weight gain did not strongly correlate with diabetes, maternal weight or the babies’ early eating habits. These findings demonstrated the issue is more complex than previously believed. However, the analysis uncovered new ideas for how doctors could more effectively talk with mothers of newborns about the issue, based on their individual backgrounds.

By working side by side, the clinicians can enhance their practices by better understanding what questions data can answer, and the ASU researchers can be directly involved in clinical research to help doctors better serve their patients.

Patton gained an even stronger personal connection to the topic as she prepared to welcome her first child during the project. She has been interested in studying practical applications in health data analysis and plans to apply what she learned in the doctoral twinning project as she starts working on her doctoral dissertation research.

In addition to helping his patients in Arizona, Coonrod presented the team’s findings to other doctors at the 2024 Pacific Coast Obstetrical and Gynecological Society Conference in Canada. Watching the ASU researchers at work also gave Coonrod the chance to practice coding skills needed to do his own data analysis in the future.

Liu says the project work has been rewarding for her as well. In her previous research focus, related to studying the molecular and genetic basis of human diseases, it would take at least 10 or 20 years to affect diagnostics and treatments. With this project, she and Patton were able to make a difference in health challenges happening right now.

“You understand what they need at this minute and how your discovery will make an immediate impact on them,” Liu says. “That's what motivated me to join this — not to develop some new AI algorithm, not to discover some new drug target, but to work with people who really need it now.”

Support from the Flinn Foundation has been critical to the project’s success so far, as this work doesn't fit neatly within the bounds of traditional grant opportunities.

“We needed a funder who is as committed to solving this complex problem as ASU and Valleywise Health are, and who was willing to work with us as we figured out how to do this in a way that is mutually beneficial for universities, hospitals and our surrounding communities in the Phoenix Bioscience Core,” says Michelle Villegas-Gold, the assistant vice president of strategic initiatives with ASU Health, a comprehensive effort to tackle the state’s urgent health care needs, now and into the future. Villegas-Gold helped create the proposal and worked with the Flinn Foundation to get it funded.

“We are grateful the Flinn Foundation and their board shared our vision for this innovative and impactful project and trusted us to carry it out with care," she says.

The collaboration between ASU and Valleywise Health continues to make impacts in clinical research.

“This partnership not only enhances care for patients but also creates a dynamic clinical learning environment where students are embedded as vital team members,” says Michael White, executive vice president and chief clinical officer at Valleywise Health. “Together, we are advancing care and sharing these transformative insights with the broader health care community.”

Liu, Nordstrom and others are setting up new doctoral twinning opportunities, exploring the use of hospital emergency department data to study heat-related illnesses as well as other research opportunities related to pediatric care.

The first round of projects — which also included research into using chest X-ray data to classify lung diseases, led by Professor Jianming Liang at ASU and Dr. Siddharth Pandya at Valleywise Health — involved the biomedical informatics program at the ASU College of Health Solutions. Future projects will also incorporate faculty members and graduate students in the behavioral health doctoral program.

Ultimately, the process developed through this partnership can be applied in other organizations and locations, advancing health care on a much larger scale.

“ASU in partnership with Valleywise Health is developing an operational model that is addressing key obstacles that normally prevent sharing and analysis of medical health records for scientific discovery,” says Mary O'Reilly, vice president of bioscience research programs at the Flinn Foundation. “The Flinn Foundation is excited to support this work as this model, once validated, can be adopted by other health partners and has the potential to unlock critical data impacting and influencing the quality of care being delivered to patients here in Arizona and beyond.”