adults airQuality athletes breathing fjell full-logo-s heartrate hjerte light lightspectrum login logo-s lydbolge movement newborn noise phone plane presence schoolchildren sleep sleepstages somnofy-quantum teenagers temperature vekter vibration vitalthings vitalthings
Back to blog
Health and welfare & Consumer & Research

Somnofy precisely classifies sleep compared to the gold standard

Somnofy seng Focus

Norway’s foremost sleep researchers have validated Somnofy against polysomnography. Learn about the results!

What is sleep?

Sleep is essential for most living organisms. Despite this, research is not clear about why we even sleep. It is though very well documented that getting enough sleep positively affects restitution, memory processing and our immune response. Too little sleep on the other hand has been connected to Alzheimer’s disease, obesity, cancer, depression, stroke and heart failure — among others. Somnofy divides sleep into 30 second longs epochs and then differentiates four sleep stages: awake, light sleep, deep sleep and REM sleep (dream sleep).

Why use Somnofy to measure sleep?

Since sleep is very important for our health, keeping an eye on the amount and quality of sleep we are getting is valuable long term information. Today’s technology is not optimal for continuous sleep measurements. The medical gold standard for sleep assessment is polysomnography (PSG), which is highly accurate, but cumbersome with many cables attached to your body sleeping in a sleep lab. This affects your sleep and is therefore not recommended for several days in a row. PSG recordings are often manually scored by a sleep physician to derive sleep stage information from your brain activity. Altogether, PSG is resource-intensive.

On the other side of the spectrum we have simpler technology like smartwatches and phone apps. These don’t affect your sleep at all or if they do, it is negligible. Sadly, this technolgy is not very accurate. EEG-based headbands measure sleep in a more valid way, but can be uncomfortable to wear at night.

Somnofy on the other hand opens the possibility to collect accurate sleep data completely contactlessly. The Somnofy unit simply sits on your night stand, or is mounted to the wall. Everything else is automatic.

How does Somnofy measure sleep?

Somnofy uses highly advanced radar technology to measure miniscule movements from your body. From the information about small movements from your body, Somnofy differentiates gross body movements from a fine breathing pattern. This data is fed to our sophisticated algorithms which have been taught sleep scoring through machine-learning. Your breathing rate during REM sleep is very different from deep sleep, for example.

Somnofy’s radar technology is completely harmless and is works even through pyjamas and blankets. If you are sharing your bed with a partner, Somnofy will not be confused by any movement which is further away than the closest breathing being — you!

The study

71 nights were measured from 71 different participants. 43 of these were women and the average age was 28.9 years (age from 19-61 years). Based on this data base, machine learning has trained the Somnofy algorithms to interpret sleep data very accurately.

• Somnofy measures breathing frequency and movement with over 99% accuracy.
• Whether the person is asleep or awake is true in 95% of cases.
• Which sleep phase the person is in hits correctly with 89% accuracy.

Conclusion: Somnofy showed high accuracy compared to medical sleep measurement (polysomnography).


The figure below shows an average precision for one night compared to PSG. For more results, please consult our publication in Sleep Medicine.

Professor Ståle Pallesen, Universitetet i Bergen

Compared to PSG (polysomnography), Somnofy is close to this gold standard for sleep measurement than any non-contact alternative.

At a glance

Measuring sleep can be important in many circumstances. Previously, it has been impractical to track sleep accurately in an easy, comfortable and cost-effective way. Somnofy is very well suited for covering some of these aspects.


Toften, Ståle., Pallesen, Ståle., Hrozanova, Maria., Moen, Frode., Grønli, Janne. (2020). Validation of sleep stage classification using non-contact radar technology and machine learning (Somnofy®). Publisert i Sleep Medicine.

Lead Data and Research Scientist

Ståle Toften

Ståle loves data and is on a mission to make your data more valuable to you. His background in physics and data science enables him to create complex models to extract valuable insight from your data, so you can sleep better and feel safer.

This website uses cookies to store statistics about the use of this website. The information is shared with Google Analytics. For more information read our Privacy Policy Agreement.