Fitbit shows wearables can detect Covid-19 early
August 26, 2020
Early findings from a Fitbit Covid-19 study suggest wearable devices can identify signs of disease at its earliest stages.
“The Covid-19 pandemic has underscored the importance of staying healthy, and our mission to help people live healthier lives has never been more important,” said Conor Heneghan, director of research into algorithms at Fitbit, in a blog post.
Since the start of this global health crisis, the Fitbit research team has been working to help make a difference in the fight against Covid-19. This includes accelerating work on early disease detection, an effort led by a team of data scientists with expertise in machine learning and predictive modelling.
In May, the company announced the launch of its Covid-19 study aimed at building an algorithm that detects the disease before symptoms start. In just over two months, more than 100,000 Fitbit users across the USA and Canada have enrolled, with more than 1000 positive cases of the virus reported.
“This study presents an exciting opportunity to see how the power of the Fitbit community will help us better understand this new and complex disease,” said Heneghan.
Because data to help detect Covid-19 are of critical global importance, Fitbit submitted its early research for publication in a peer-reviewed journal. While it works to finalise the publication, it has made the full manuscript publicly available as a preprint so it can share some of the preliminary findings.
“So far, we’re encouraged to see physiological signs of disease detected by Fitbit devices simultaneously with study participants’ reporting the onset of Covid-19 symptoms, and in some cases even before,” said Heneghan. “Based on the findings of our study, we can detect nearly 50 per cent of Covid-19 cases one day before participants reported the onset of symptoms with 70 per cent specificity.”
This is important because people can transmit the virus before they realise they have symptoms or when they have no symptoms at all. If people know they should get tested a day before symptoms begin, they can isolate and seek care sooner, helping reduce the spread of Covid-19.
“As researchers, we are always working to find a balance between sensitivity – alerting people who may be sick – and specificity – the ability to identify people who are healthy – as there are trade-offs to both,” said Heneghan. “We will continue to work with the clinical and public health communities to evaluate different models for developing this technology to ensure the optimal balance.”
The study also reinforces that breathing rate, resting heart rate and heart rate variability (HRV) are all useful metrics for indicating onset of illness and are best tracked at night, when the body is at rest. The research shows that HRV, which is the beat-to-beat variation of the heart, often decreases in people who are exhibiting symptoms of illness, while resting heart rate and breathing rate are often elevated. In some cases, those metrics begin to signal changes nearly a week before participants reported symptoms.
On average, HRV hits its lowest point the day after symptoms are reported. Increases in resting heart rate normalise, on average, at least five to seven days after the start of symptoms. Breathing rate peaks typically on day two of symptoms, but there is a slight elevation, on average, for up to three weeks after symptoms start
In addition to detecting early signals of Covid-19, the researchers are also gleaning some insights into common symptoms, as well as severity, duration of illness and the symptoms most likely linked to hospitalisation.
“Many of these observations align with what we’re hearing from other researchers and public health officials,” said Heneghan. “For example, being older, male or having a high BMI increases the likelihood of severe outcomes.”
In addition, the study found that shortness of breath and vomiting are the symptoms most likely to predict that someone with Covid-19 will need to be hospitalised, while sore throat and stomach ache were the symptoms least likely to predict the need for hospitalisation.
The most common symptom reported by individuals with Covid-19 is fatigue, which was present in 72 per cent of participants reporting having Covid-19. This was followed by headache (65 per cent), body ache (63 per cent), decrease in taste and smell (60 per cent) and cough (59 per cent). Fever was present in just 55 per cent of people reporting Covid-19, an indicator that temperature screening alone may not be enough to understand who might be infected.
Mild cases – those who recovered at home on their own – show a median duration of eight days, while moderate cases – those who recovered at home with help from others – last about a week longer, with a median duration of 15 days. For severe cases as inpatients who end up requiring hospitalisation, the median duration of illness was approximately 24 days. But this duration had a large spread, with several cases lasting longer than two months.
“It’s clear that our bodies start to signal impacts from the disease before more noticeable symptoms appear,” said Heneghan. “With these initial signals identified, we’ll continue our work in developing an algorithm to detect diseases like Covid-19 and focus on expanded research in a real-world environment.
“Early detection is critical, and we hope to bring this type of information to consumers as soon as possible. As a next step, we’ll continue to work with our research partners like Scripps Research Translational Institute and Stanford Healthcare Innovation Lab to further validate the technology and intend to engage with the appropriate regulators globally to determine the best path to bring this to consumers.”