Duke scientists are studying if your watch could know if you're sick before you do

Researchers at Duke University are currently in the enrollment phase of a new study to determine if your smartwatch could hold the key to reducing the spread of respiratory illnesses.

Study lead Jessilyn Dunn explained researchers with the university’s BIG IDEAS LAB have reached out to over 350,000 people for the study.

Everyone they’ve contacted was tested through Duke Health for either COVID-19, the flu, or RSV between 2018-2023. Data from a participant’s smartwatch is then being analyzed during the time when they were sick.

“The goal is to try to build a tool that can detect early signs of infection before a person might even have any symptoms of being sick,” said Dunn. “The idea would be that you could prevent somebody from going out and spreading infection, unknowingly.”

Scientists are currently looking at those who wear commercial smartwatches like an Apple Watch, Samsung Watch, Garmin Watch or Fitbit. Data of interest includes heart rate, movement, sleep, temperature, and oxygen levels.

“We build machine learning algorithms that take all of that information together, and predict whether or not somebody has an infection,” explained Dunn.

She continued, “The beauty of this current study is that we're actually able to pull data retrospectively. So what we have is the ability to look back into the electronic health record and say, ‘This person actually tested positive for COVID-19, RSV, flu on this date.” We know we have this really sort of ‘clean data’ that tells us that, ‘Yes, somebody was sick at this date and time.’”

Dunn said the study looks at adult patients from all different demographics.

“One of the main targets of this study is to make sure that we have a balanced demographic, such that any algorithm that we build works for everyone,” said Dunn.

The researcher explained the current study is built on previous work she and others have been doing for over five years.

“When the pandemic came around, it became obvious that the methods that we've been developing to detect infection from common consumer smartwatches would be very applicable for these respiratory infections that are spreading rapidly,” said Dunn. “I think the foundation of the technologies that we had been developing for the previous five or so years before the pandemic, combined with the huge amount of data that became available, made it clear that this is the time to do this study.”

Dunn said the researchers are working with BARDA, a U.S. Department of Health and Human Services office focused on the development of medical countermeasures.

The study is expected to take 24 months to complete.