Jessilyn Dunn: Gaining Insights from Biomedical Big Data

June 5, 2018

New faculty member Jessilyn Dunn explores how to use biomedical data sets to model and guide personalized therapies

Jessilyn Dunn

Jessilyn Dunn

Jessilyn Dunn will join the faculty of Duke University’s Department of Biomedical Engineering (BME) and Department of Biostatistics & Bioinformatics (B&B) beginning January 1, 2019. With research centered on using large-scale biomedical data sets to determine positive health outcomes, Dunn will contribute to the growing focus on biomedical and health data sciences at Duke.

Today, people are most familiar with health data due to the rising popularity of wearable technology, like fitness trackers or smart watches. These tools, which measure daily activity, workouts and sleep patterns, provide users with a general idea about their overall health and help them to form healthier lifestyle habits. Physicians use information that they collect from medical records and clinical chemistry tests, which identify biomarkers for complex diseases or disorders, to decide on the optimal treatment plans for patients. Large biomolecular data sets like genomics, proteomics and metabolomics are used largely by biomedical researchers, and their role in healthcare is expected to grow in the future as their clinical value is demonstrated. There has been little to no integration across these diverse types of biomedical data for research or healthcare to date.

In her new role at Duke, Dunn aims to draw novel health insights and actionable conclusions by integrating data from these three sources (wearable devices, electronic health records, and multi-omics). A major focus of the Dunn Lab will be generating predictive models for different diseases using statistical and machine learning methods with the ultimate goal of developing effective and customized treatment plans tailored to individual patients.

“I’m excited to bring a new facet of health and data science to Duke BME and B&B,” said Dunn. “By integrating diverse types of biomedical data, we can develop more precise ways to catch who is sick and who will become sick, and then develop appropriate therapeutic interventions. My goal is to turn data sources, like wearable technology, from an information delivery system into an insight delivery system to help people lead healthier lives.”

Although data science is useful across a broad spectrum of diseases, Dunn has focused her efforts on two areas: cardiometabolic disease and pre-term birth and pregnancy.

Cardiometabolic disease is a combination of disorders that includes both cardiovascular conditions, like heart disease, and metabolic diseases, like type 2 diabetes. Because many of these disorders share common symptoms, physicians typically use similar therapeutic treatment plans to treat a variety of cardiometabolic disorders. Through her work, Dunn hopes to use patient data to better inform and customize treatment plans.

“We’re still trying to work out what kind of data is the most useful for creating a treatment plan for cardiometabolic disorders,” said Dunn. “Consumer wearables hold promise because we’re able to collect high-resolution data over long periods of time. That constant stream of data may be more useful when monitoring and treating cardiometabolic disease than blood tests and vital-sign testing that only occur infrequently at doctor’s visits.”

In the realm of pregnancy, Dunn is using biomedical data to predict and prevent dangerous issues like preterm birth, which affects children born before the 37th week of pregnancy. Each year, more than one million newborns die from complications relating to preterm birth, and children who survive are more likely to develop health complications like breathing problems or developmental delays.

“There are lots of problems that can occur during pregnancy that we don’t fully understand, and because of that we don’t always have effective methods for early intervention,” said Dunn. “One of the ways we’ve tried to incorporate data involved a wearable device to monitor EMG signals from the uterine wall. This type of data could help us develop computational models that are useful for predicting complications.”

Dunn, who is currently a postdoctoral fellow at Stanford University, obtained a PhD in BME from Georgia Tech and Emory and a BS in BME from Johns Hopkins University. During that time, she studied the transcriptome and the epigenome of certain types of vascular cells to understand how these cells changed during disease processes. This research, she says, ignited a strong interest in the potential of data science, especially surrounding the potential of wearable technology to detect disease at different biological scales.

“It was always exciting to see how our findings changed as we layered in different types of data from multiple sources,” said Dunn. “When I started at Stanford, I was thinking about what data we could add to information in medical records and biomolecular data sets that would have an impact. Wearable technology was the obvious next step.”

Now, as Dunn prepares to make the transition to Duke, she looks forward to growing the field of data science with collaborators across the university.

“Duke is at an exciting transition point, where there is a wide interest in using data science to revolutionize healthcare,” said Dunn. “I’m excited to build up the data science community within BME and drive Duke’s reputation as a premier institution in this fascinating and growing field.”