Sleep tracking just got a significant upgrade for public preview users

[StefanieFitbit]

Hey Fitbit Community,

We know sleep is about more than just time in bed; it’s the complete story of your night. By distinguishing between when you’re truly aiming for sleep and when you’re just relaxing, the Coach can more accurately guide your next move. 

This approach is why, starting last year, we’ve revamped our sleep tracking using our latest advances in sleep science and machine learning to significantly improve its accuracy first for public preview users, and later for everyone 

We’re deploying more advanced machine learning to update our algorithm so we can better capture when you’re trying to sleep, recognize when your sleep is interrupted, and sharpen our detection of those critical stages like REM, Light, and Deep. With these newest improvements we’re seeing a +15% improvement in our sleep tracking accuracy(1).

Capturing the complete picture of your sleep

Tracking your sleep is about more than just logging sleep stages. To truly capture your entire sleep experience, we expanded the scope of how our algorithms estimate when you were intending to sleep.They now do a better job of telling if you’re actually aiming for sleep or just relaxing with a book, giving a more accurate picture of how you’re utilizing your time in bed. 

The systems are also sharper at identifying middle-of-the-night interruptions—like a fussy pet or a trip to the kitchen—versus general restlessness. We’ve also enhanced our detection of naps that happen outside of your usual bedtime schedule, capturing more of your rest throughout the day.

Built for every sleeper

In addition to more advanced machine learning models, we expanded them with thousands of nights of data. To make sure our sleep experience works for a broader diversity of sleepers, we trained our models on larger, more inclusive datasets encompassing more people from more walks of life and a greater variety of sleep issues, collected from a broader range of environments—both in and out of the lab.

"Recognizing the wide variety of sleeper types out there, our team worked diligently to improve our algorithms' ability to characterize sleep for everyone. These improvements range from better estimates of how your sensor data reflects your sleep-wake behaviors, to a more accurate staging algorithm based on cutting-edge machine learning from a more diverse group of people, to a uniquely user-centric interpretation of sleep and wake periods.

Guided by regular feedback from an advisory panel of sleep and health experts and our extensive research—both with users and in scientific datasets—we're rolling out these changes in order to enhance how people understand and can work to improve their sleep health."

- Logan Schneider, Adjunct Clinical Assoc Prof of Sleep Medicine @ Stanford
Neurologist @ Stanford/VA Alzheimer's Research Center
Clinical Lead for Sleep Health - Google

For a deeper dive

Curious about the data behind these improvements? [Check out our white paper] for an in-depth look at how we evaluated our latest sleep-tracking algorithms.

What's next?

To help you make sense of your sleep, we are working on a reimagined Sleep Score designed to be more actionable and transparent.  The new score evaluates specific aspects of your rest – not only how much sound sleep you got, but also how long it took you to get there. This level of detail provides targeted coaching, helping you understand exactly where to focus your routine for better recovery.

The updates to the sleep tracking will be rolled out over the next few days and the new sleep score experience will be available over the next few weeks. This will only be rolled out to Public Preview users.

Thank you,
Fitbit Community

Disclaimer 
(1) Improvement in sleep stage accuracy based on clinical validation of sleep algorithm against gold-standard measurements, comparing our latest sleep tracking algorithm to our previous generation across all compatible Pixel and Fitbit devices.