Sleep tracking is a tricky one. Almost all of our wearables claim they can do it, but try slapping on a bunch of fitness trackers and checking the results in the morning – we guarantee you'll see a lot of disagreement. Much like calories, it's a science that people have been slower at getting right, which is kind of understandable when the current "gold standard" is considered to be the confines of a sleep laboratory. What's more, there are loads of factors that play into sleep, so it's hard for tech to accurately monitor how much good sleep you're getting, let alone diagnose the cause of bad sleep.
Fitbit made a big push in this field earlier this year with the launch of its sleep stages and insights features, which for the first time in the company's history gave users a meaningful look into their nightly activity and what all the oft-confusing numbers meant. Now, after working with independent experts on some studies, which put its trackers against hardcore lab equipment, Fitbit claims it has validated the accuracy of wrist-worn trackers in determining our sleep stages.
The results were just published in the latest findings by the Sleep Research Society. Fitbit's results came in at a 69% agreement with testing by polysomnography technicians (essentially the best way to track sleep) in "normal adult sleepers", which is far from perfect, but considered a "reasonable degree of accuracy" by the study – and better than previous non-EEG sleep staging done with similar tech outside of a lab.
"We've basically exceeded anything that has been published in academic literature about how you can measure sleep stages with respiration, heart rate, movement and so on," Dr Conor Heneghan, Fitbit's lead sleep research scientist, told Wareable. Fitbit studied participants with the older Surge tracker, but the same heart rate and accelerometer tech is found in Fitbit's more recent devices. These results are good news for everyone else though, as while Fitbit has its own proprietary tech, combining heart rate and movement is something other trackers already do, even if they don't yet use it for separating sleep stages.
Rythm is trying to bring lab-quality sleep tracking into the bedroom
Bringing that gold standard out of the lab and into our homes is something being worked on by companies like Rythm and EverSleep, and this new study could be an important waypoint in the journey. Longitudinal sleep studies are currently difficult, often impossible, as you'd need to be in a sleep lab for a long period of time to get really meaningful information. But if we can do it at home in our own beds, with a bit of tech wrapped around our arms, suddenly you're getting insights that nobody has ever had before. Hopefully this validation study will nudge more wearable companies to start taking sleep a bit more seriously. Heart rate is the key here, as accelerometer data alone can't really distinguish REM from light or deep sleep – despite what your tracker might tell you.
Heneghan will be presenting the study at the SLEEP 2017 conference, along with some findings that Fitbit took from its database of 4 billion nights of sleep data. While there was nothing hugely revolutionary in the numbers, Fitbit did pick out a few interesting points. "It's a little bit early to fully understand but we're seeing that people who have short sleep are getting a little less deep sleep than would have perhaps been predicted by the current understanding," said Heneghan. The traditional view has been that the body prioritises deep sleep, but Heneghan said Fitbit's studies suggest thinking might have been overly generous. "We're kind of seeing evidence that the body isn't quite as efficient as it could be. Maybe a short sleep isn't even getting as much deep sleep as we would like them to get."
He also said they had noticed that people who wake up a bit earlier are losing a little bit of REM sleep compared to the general population, which makes sense as more REM sleep tends to happen later in the night.
The higher the resolution of sleep data, the better the insights can be, and Fitbit says that while this is all an exploratory look for now it's going to start integrating its findings more into the user experience. So the little messages that pop up in the app (if you're using Sleep Insights) should get more useful over time. And the more information our trackers collect, the more our devices can start painting the bigger picture. Are those late-night yoga sessions helping you sleep better; is running in the evening having a negative impact? "We suddenly allow a lot of scientific work to happen on a consumer level. People can explore their own sleep patterns in much more detail and fidelity than they could before."
Alongside this, many myths and misunderstandings about sleep can be shattered too. For example, you don't actually need as much deep sleep as you might think – about 15% is actually fine! – and waking up several times in the night is normal as well. "People sometimes aren't aware of the fact it's quite normal to wake up quite a lot in the night," said Heneghan. "You don't usually recall it, if you have 30 minutes a night it's normal physiologically."
So, what happens now? "I kind of suspect what will happen is that other researchers will pick up [the research] independently," says Heneghan. "My guess is that this will happen over the next year or so. And the scientific acceptance will grow as those studies emerge. Obviously people want to verify it themselves."
Fitbit's findings aren't going to change the field of understanding overnight, but it's hopefully another small step towards fully understanding what goes on in those mysterious hours of shut-eye.
How we test