Fitbit data suggest the potential for early COVID-19 detection using wearables

• Fitbit has released data on its efforts to develop an algorithm to detect COVID-19 using a device that was recorded before the signal started.

• In the journal Nature and Digital Medicine, researchers showed changes such as respiratory rate and heart rate in the days before respiratory symptoms appeared. With a 90% scan, the algorithm can detect 21% of the study the day before symptoms appear.

• The algorithm was not implemented in the last analysis. Fitbit is working with Northwell Health to identify the first tests of the COVID-19 algorithm.

Attempts to use data from wearable devices like Fitbit devices and infectious disease monitoring results in coronavirus infection as well as by the Scripps Research Translational Institute and other places where the results of their previous research on the concept of the disease are published. However, COVID-19 is of great interest as public health experts are looking for ways to stop the transmission of viruses.

In theory, changes in vital signs that occur between viral infections and the onset of symptoms can aid in the early detection of COVID-19 so that individuals can take steps to prevent further infections, but if treatment is good. The situation is getting worse and worse.

As a device that can provide data to support such predictions, Fitbit began enrolling device users in a study earlier this year before the first results were released in August. The company released its full report on Monday.

The article describes the development of an algorithm based on data from 2,745 North American Fitbit users identified for COVID-19. These people’s heart rate, and especially the number of breaths, increases in the days before symptoms appear. Fitbit used data and developed an algorithm to determine if a person had COVID-19.

With an assumption of 90%, the algorithm recognized 43% of the test day after the onset of symptoms and 21% of the test day before the onset of symptoms. An increased concentration reduces the research risk. There was a 99% chance that the algorithm detected 7% of the test days before symptoms appeared.

Fitbit used data during the peak of the COVID-19 explosion in New York State to make it clear that the algorithm could work in a larger explosion. The researchers estimate that the daily spread of the disease in the state is 0.65% in mid-April. In this case, with 99%, about a seventh of the prediction algorithm would be very good. However, the accuracy is lower in the region with the lowest case.

Although the algorithm produces far more falsehoods than the real ones, even with large numbers, the researchers say it can encourage people to be careful or experimental. The spread of this virus can be slowed if think tanks need to be isolated for self-defense when the algorithm warns them. With a specification of 99%, the algorithm can still accurately predict that a person is not infected.

Along with the results from Fitbit, there is a lot of information, including how employees remember when symptoms started. Studies are needed to determine whether the algorithm can predict whether a person has COVID-19. Fitbit, sponsored by the Department of Defense, plans to test this type of test. Other organizations, including Sonica Health, have received government funding for wearable coronavirus devices.