Improving Patient Outcomes by Analyzing Wearable Data
By Diane Chernoff
MEDx Investigator Jessilyn Dunn brings fresh perspective on big data and wearable technology to Duke’s growing network of engineering and medical collaborations
Over the past few years, Duke University has launched an interdisciplinary initiative to strengthen research and innovation in the realms of data science, engineering and medicine.
MEDx (Medicine and Engineering at Duke, https://medx.duke.edu/), a collaboration between the Pratt School of Engineering and School of Medicine begun in summer 2015, promotes the exchange of ideas between doctors, engineers, researchers and innovators. Through interdisciplinary programs, MEDx hopes to stimulate the development of new therapies, diagnostics, devices and educational opportunities in order to accelerate science and its translation to clinical practice to improve patient care.
The effort is already bearing fruit, as Duke has proudly made advancements in health informatics, drug and device innovation, tissue and genetic engineering, reverse-engineering the brain, and basic science fields. One of the primary areas of growth has been driven by new biomedical imaging and sensing technologies that generate new data for electronic health records and beyond. By studying these topics, students and faculty seek to develop innovative data science, machine learning and digital health modeling approaches to improve biomedical data analytics and patient care.
To build on its leadership in this arena, Duke University’s Departments of Biomedical Engineering and Biostatistics and Bioinformatics welcomed Jessilyn Dunn as a new faculty member and MEDx investigator (with appointments in both engineering and medicine) in January 2019. Dunn served as a postdoctoral fellow at Stanford University, obtained a PhD in biomedical engineering from Georgia Tech and Emory University, and graduated from Johns Hopkins University with a BSE in biomedical engineering. In previous work, Dunn has conducted research in multi-modal biomedical data integration with the goal of improving personalized medicine.
Each year, MEDx works alongside departments in both the School of Medicine and Pratt School of Engineering to award grants to support pilot projects. This year, one of the winning projects is spearheaded by Dunn and Mark Feinglos, professor of medicine and associate professor of psychiatry and behavior sciences and pathology. Together, the two are examining digital biomarkers of pre-diabetes and glycemic variability. The project coincides with Dunn’s research on multi-omic biomolecular data sets, which include genomics, epigenomics, proteomics and metabolomics.
In addition to her MEDx project, Dunn plans to continue her research in biomedical big data at Duke, with a focus on studying the information stored in consumer wearable devices to gain insights into patient health.
As medical technology becomes more accessible to patients outside the clinic—as well as more more ubiquitous and consumer friendly—Dunn has learned to take advantage of such technologies. She has studied data collected from wearable devices like Apple Watches and Fitbits with the intention of using the data to make general predictions on patient conditions that can be applied to larger patient populations and improve future patient care.
Many wearable devices on the market today collect data on users’ daily activity, sleep patterns, workout habits and basic conditions like heart rate and body temperature. “Most of life occurs outside of the clinic, and by analyzing the data recorded by these wearable devices, we can capture patient health outside of the clinic,” Dunn explained.
Dunn hopes to collect information that is clinically actionable from the wearable devices, like the users’ heart rate, physical activity, temperature, electrodermal activity and location, to make general predictions on patient health. “We can use the data to make a prediction that someone is deteriorating in some area based on the data we can collect from their watch,” Dunn said.
Dunn’s research is not case specific, and she hopes to apply large-scale data science principles on data collected from wearable devices to broader populations. “Instead of focusing on developing more complex sensors, we hope to understand what more basic sensors can accomplish and translate the things that we learn from individual patients to a broader population as well as make predictions about future health events,” Dunn added.
Dunn’s research is focused on studying data collected from wearable devices to make improvements in treating patients with cardiometabolic disease. Cardiometabolic diseases, which are combinations of cardiovascular diseases, like heart disease, and metabolic conditions, such as diabetes, are epidemics across the country.
As Dunn joins Duke, she looks forward to working with patients, physicians and researchers who may benefit and contribute to her research, saying, “We are developing a comprehensive infrastructure for personalized risk classification and tailored, remote intervention strategies.”
In the coming months, Dunn is looking forward to taking advantage of Duke’s “stateof- the-art medical facilities, data science and localization to the Southeast.” Through Duke’s cross-campus initiatives to integrate biomedical engineering and medical research, she is confident that her research will accelerate through collaborations with MEDx and other research and clinical groups on campus.
As Dunn prepares to join the Duke family, she says she is “most looking forward to getting involved with both the Duke and surrounding community and building an impactful translational research program.”
Diane Chernoff is a first-year Pratt student planning to major in biomedical engineering from New York.