After two months of study with more than 100,000 Fitbit users in the trial and 1000+ positive COVID-19 virus reports, Fitbit has released preliminary results from its early disease detection work. The wearables company experts are employing machine learning and predictive modeling to build an algorithm that can signal the disease before users develop symptoms. The study’s preliminary draft in preprint form — not yet peer-reviewed — contains the early findings.
So far, the researchers report the study algorithms successfully detect with almost 50% sensitivity COVID-19 cases one day before the users have reported symptoms, meaning they are correct in positive virus findings almost half the time. The early results also have 70% specificity, which means they are accurate 70% of the time identifying people without the virus. The study uses respiration rate, heart rate, and heart rate variability to detect the disease. The readings are most useful when they are taken while the user is resting at night. Specifically, the researchers report heart rate and breathing rate elevate and heart rate variability (HRV) decreases up to a week before users notice and report symptoms.
There are many additional findings in the preliminary report. We look forward to a final study report, because the sooner early detection tests are validated and approved, the faster health care workers and individuals will be able to predict the probability of the virus occurring. The result will be faster treatment and potentially less exposure to other people.
The Fitbit study is ongoing. If you’d like to participate in the study, you can join on the Assessments & Discover tab in the Fitbit app or via this link.
Trackbacks/Pingbacks