Oura study shows covid-19 can impact your biometrics for weeks after infection

At the start of the pandemic, researchers and wearables makers rushed to see whether smartwatches and fitness trackers could detect covid-19. It’s now 2023, and while wearables show promise in detecting illnesses, there hasn’t been too much progress on the covid front. But even as most people start moving on with their lives, some wearable makers are still sifting through the data to see what can be learned from the past three years. Case in point: smart ring maker Oura just released a new study that found significant changes in its users’ biometrics up to 2.5 days before and 10 days after users reported a covid-19 infection.

The study, which was published in the peer-reviewed journal Digital Biomarkers, looked at 838 Oura members who had self-reported covid-19 infections as well as 20,267 members who had received a covid vaccine. It compared temperature, resting heart rate, heart rate variability, breathing rate, and sleep efficiency in the month before and after a vaccination or infection. It also compared the physiological responses between variants, age groups, and vaccination status.

The findings pretty much corroborate the existing literature when it comes to covid-19 detection. For example, the bodily response to the delta variants was much stronger than earlier variants. Meanwhile, getting vaccinated produced a similar but less severe response than an actual infection, lasting up to four days after the shot. In the same vein, vaccinated users had a less severe reaction than unvaccinated users. The response was also found to be greater in users under 35 years of age compared to those over 50.

“We were somewhat surprised to see responses in all of the biometrics that we looked at,” Oura head of data science Hal Tily told The Verge. Tily went on to note that, in this particular study, Oura was able to document the course of infection in more detail than it had previously. For example, Tily said that while temperature differences tended to return to baselines after about 10 days, breathing rate and heart rate changes could continue for as long as 20 days. While this can’t be taken as evidence of long covid, it does suggest that it takes some people more time to return to their baseline biometric levels after infection.

“On average, we could see symptoms pretty clearly extending days, if not weeks, and in some cases as long as a month following infection,” Tily said.

This study does have its limitations, however. It may have been published in a peer-reviewed journal, but that doesn’t make it a clinical study.

To begin with, no volunteers were recruited, and it wasn’t a double-blind, randomized trial, which is the gold standard in clinical research. According to Tily, the data was pulled from existing Oura users in accordance with its terms of use and privacy policy, which grants Oura the ability to use anonymized, aggregated data to develop new features or for research. The company also identified which users had gotten covid or vaccinations based on self-reported tags that users can use to add context to data from a certain day.

Because Oura only looked at aggregated averages for the data, Tily says it didn’t require users to opt in as a more rigorous study involving individualized data would — such as the one Oura did with the University of California San Francisco on temperature and covid-19. (Working with accumulated anonymized user data is a common practice among wearable companies. If you’re an Oura user and that creeps you out, Oura does allow you to obtain and delete your own data under its privacy policy.)

Tily admitted these limitations when asked by The Verge, explaining that the goal of the study wasn’t to provide any kind of diagnosis, draw definitive conclusions, or even pass FDA regulations.

“My real hope is that we get more people excited about the possibility of using commercial wearable data to do large-scale monitoring of illness,” said Tily, noting that the benefit of consumer wearables is the sheer volume of data they generate, which can be useful in identifying large-scale trends. For example, instead of focusing on warning individuals that they could be falling ill, consumer wearables may be more helpful in passively monitoring potential outbreaks in a particular region to inform public health policies.

However, not everyone in the scientific community is keen to trust data from commercial devices. Compared to consumer tech, medical devices undergo stricter levels of calibration, sanitization, and data privacy. Consumer wearables are also more limited by their user bases. In this case, Oura’s dataset may not accurately reflect certain ethnic or socioeconomic groups simply because it’s a niche device that appeals to self-quantification nerds who can afford a $300 device with a $6 monthly subscription.

But even if scientists were to get fully on board, I wouldn’t hold my breath for the possibility that consumer wearables will be predicting illnesses anytime soon. Studies like this demonstrate that wearables can detect physiological changes after infection, but the next step is figuring out whether these devices can tell the flu from covid. After that, there are a host of other factors to consider, like clinical validation, the regulatory process, and whether this is something that’ll be gatekept to specific wearable ecosystems.

Essentially, the point of this kind of research isn’t to figure out how to handle covid-19 as we know it. It’s to possibly get a jump start on the next pandemic.

Photography by Victoria Song / The Verge