Since we launched Sleep Gadgets two years ago we haven’t really paid that much attention to sleep tracking technology. We have our reasons…. First, the sheer number of devices that claim to be able to monitor, measure and analyse your sleep is overwhelming – from generic wearables that cost less than $5, to smart beds and mattresses costing 10s of thousands, there are innumerable devices that purport to be able to give you insights into your sleep -and we just don’t have the resources, nor the motivation to cover the length and breadth of the market in detail.
But perhaps more importantly, in recent years, scientists and medical practitioners have become increasingly wary of recommending sleep tracking technology to those who suffer sleep problems. The reason is two-fold; first, very few consumer sleep trackers have undergone rigorous scientific validation against the ‘gold standard’ of sleep measuring tools, polysomnography (PSG), used by medical professionals and researchers in diagnosing, treating and studying sleep disorders.
Secondly, it’s widely understood that using a tracker might actually make your sleep problems worse. In recent years, the reduced cost, availability, and often dubious marketing claims have led many individuals with sleep anxieties to turn to consumer sleep trackers to understand and potentially fix their sleep problems. But in many cases, this turns into an unhealthy obsession with achieving the perfect sleep score – which obviously is not a recipe for getting rid of your sleep anxieties.
But despite these negatives, there are still a plethora of applications for a consumer product that can accurately monitor your bio-signs whilst you’re asleep, and one of the most exciting such devices is Somnofy from Norwegian health-tech startup VitalThings.
In development since 2017, Somnofy is a bedside device that uses ultra-low power radio waves to detect body movement, which it then crunches via AI into a detailed analysis of your sleep, including sleep stages, respiratory rate and a whole lot more.
Whilst we’ve seen this kind of tech before, what’s impressive about Somnofy is that the device has undergone rigorous clinical validation against PSG, as well as offering a host of innovative, connected features including a smart alarm, environmental monitoring, guided breathing functions, and even the ability to function as an ‘early warning system’ against illness – something that has become more and more relevant in an age of Covid-19.
Somnofy is the first sleep tracker we’ve seen for a long time that we’re really excited about, so we spoke to Alf-Egil Bogen, one of the two founders of VitalThings, who answered our questions about Somnofy.
Jeff Mann (JM): Can you give us a brief ‘elevator pitch’ for Somnofy?
Alf-Egil Bogen (AB): Somnofy is a contactless sleep assistant that can do complete sleep analysis, complete bedroom analysis, and provides an active coach on how you can improve your sleep based on your individual data.
It also provides active help to fall asleep through closed loop guided breathing to actively help you relax and hopefully fall asleep.
Somnofy also has a smart alarm that calculates your ideal wake time (within a set timeframe) to avoid waking you from deep sleep. This makes you feel much more refreshed when you wake up.
Somnofy also measures your breathing rate to establish your personal baseline and to show you if there are any changes that are significant.
JM: What was the inspiration and motivation for developing Somnofy?
AB: The inspiration came from an advanced baby monitor we developed earlier. When we tested it we found that both me and my co-founder had sleep apnea without having an idea about that earlier.
We then started to read about sleep and realised how many people had various types of sleep problems. We started to talk to sleep scientists and they basically said that none of the wearables or gadgets were useful for high accuracy sleep monitoring.
So we decided to develop Somnofy and let the experts train our machine learning algorithms to see how close we could get to the medical gold standard polysomnography (PSG).
The validation study was done by the scientists at Bergen University in Norway and published in Sleep Medicine. Many sleep scientists are already now using Somnofy and we are now gradually entering the consumer market with the same high precision technology.
JM: Who is Somnofy designed for and what are the advantages of using Somnofy over, let’s say a Fitbit or an Apple Watch for sleep tracking?
AB: In general wearables have show an accuracy of between 50-60 % of PSG. With Somnofy you get between 80 and 90% which is very useful for science applications.
Apart from this major advantage in accuracy, with Somnofy don’t need anything attached to your body. Some people don’t like to sleep with a watch. Somnofy is not a wearable, it sits on your night stand and analyzes sleep completely without any body contact. Some experts call these devices ‘nearables’.
Another application is the health and welfare market. In Norway we are established partners for healthcare institutions and the research community. With Somnofy Live Monitor, elderly homes benefit from a night shift prioritization tool, plus sleep researchers get a tool for large scale sleep data in populations that used to be difficult to study.
We have also been working with elite athletes and the Norwegian Olympic team for more than two year now. They have used Somnofy to see how training and stress will influence sleep, and how sleep optimization can increase their performance. This knowledge, in addition to the experience we gain from working with sleep scientists has formed the Somnofy we now have launched to the consumer market.
If you want to make any changes in your life you need to be motivated. And sometimes you can get more motivated as you learn more. Somnofy will help you learn much more about your sleep and sleep hygiene and can be a tool to help you make changes.
Improving sleep will give you more energy to do things in the day, reduce risk of injuries or accidents, help some people maintain a healthy weight, reduce depression and last, but not least, reduce the risk of long term sleep related chronic diseases.
Somnofy gives you individual guidance through the Sleep Coach within the Somnofy App. The Sleep Coach uses your personal and environmental data and over time can help you understand the specific factors that influence your sleep.
JM: Tell us more about your work with the sports and athletic communities.
AB: We started early to work with the Norwegian Olympic team. Elite athletes need to work in great detail on all aspects of their sport – nutrition, sleep and life in general – to perform at their best. Their lives are very structured and they have to do a lot of tests and measurements to document the effect of changes.
Using Somnofy to learn how sleep can optimize health and performance was therefore very useful. Somnofy has been used in several research programs as well as on an individual level to find the balance between training, restitution and sleep. The Norweagian soccer team also used Somnofy during the World Championship to optimize room placement during the competition.
Many aspects of what we learned from elite athletes are implemented in the consumer version of Somnofy.
JM: We’ve seen contactless sleep sensors before – Resmed, SleepScore, apps etc. How does Somnofy differ from these previous technologies?
AB: The biggest differentiator is still our accuracy. Compared to Resmed and SleepScore, we achieve higher agreement with PSG and our validation study published in Sleep Magazine is second to none.
The underlying detection technology is similar to Resmed, but also very different. In Somnofy we use the most advanced chip solutions available for detecting small movements from a distance. We developed the chip before we started VitalThings.
We can separate movements by distance so two people in the same bed is not a problem. Somnofy transmits extremely low power, about 1/1000 the power of a Bluetooth handsfree set such as Airpods.
And with Somnofy you never have to start or stop anything, neither on the device, nor in the app. There are algorithms that detect when you go to bed and when you intend to sleep.
JM: Sleep trackers have taken a lot of flack in recent years – some say they can cause additional anxiety (orthosomnia) if you have a sleep problem. How would you speak to these criticisms?
AB: The criticism is accurate. Incorrect use of any sleep technology has the potential to exacerbate underlying sleep problems. Especially if the accuracy of the tracker is insufficient. The scientists we have been working with say that a low accuracy sleep tracker can cause sleep problems. That is why we worked so hard to achieve the highest accuracy in the market to day on any contact less sleep tracker. Sleep scientists are using Somnofy in their own projects which is a good indication of the quality.
We worked intensively with medical experts in developing Somnofy and we also have two medical doctors on our team. We are confident and have proof of the fact that Somnofy’s measurements actually show happens during sleep. We state the facts, so to speak. What you do with the facts is in part up to you. We do deliver guidance for better sleep through our built-in sleep coach which can be accessed through our Somnofy App.
And that is the second point: with proper guidance and understanding of the sleep analyses we provide through our sleep guide, our customers become empowered with the knowledge, not overwhelmed with fancy graphs, that lack practical tips.
JM: Another criticism of sleep tracking technology is that most devices are unvalidated against clinical standards. Can you tell us about the recent validation study of Somnofy and its results.
AB: The gold standard in classifying sleep stages is the use of polysomnography (PSG). This is used around the world to identify sleep disorders and to classify sleep stages down to 30seconds intervals.
In our classification we used 71 people where some slept at home and others slept in a sleep lab. In the sleep lab we had two Somnofy devices in different positions in combination with PSG with video recording. For the people at home we had one Somnofy and full PSG without video. The datasets were scored by 5 non-affiliated sleep scoring experts and we used the results to do a full in depth statistical analysis of the match between the two methods. This is one of the largest validation studies we are aware of.
We had an almost 90% match with the Cohen’s Kappa analysis and higher accuracy than any other contactless device. Even at level with some of the EEG and PSG validations we found online. We are extremely accurate not only in determining if a person is sleeping or only resting, but we also differentiate well between light sleep, deep sleep and REM. These findings correspond, or rather build on, the 99% accuracy in tracking breathing rate and movement of the body.
JM: Somnofy uses AI to make predictions about sleep behaviours. Can you explain the fundamental differences, pros and cons of measuring sleep using PSG, and using an approach like Somnofy?
AB: Polysomnography measures EEG, ECG, EMG, airflow, and more, to provide a complete picture of night sleep at a sleep laboratory. This data is collected and manually scored in 5 sleep phases by trained personnel. With myriads of cables attached all over your body and head, face, many people struggle to sleep as they do at home in their own bed.
This method has been developed in the 1970s and has changed little since. It is the most studied method for accurate sleep analysis and agreement between two different scorers is high (80-95%). As the medical gold standard, this method is considered to provide the final answer regarding sleep quality and PSG is used to diagnose severe sleep disorders.
Somnofy uses a very accurate movement sensor and measure breathing rate and movement on the body. Based on the neural network that powers the machine learning capabilities of Somnofy, our sleep monitor was able to ‘learn’ what type of movement and breathing rate corresponds to which sleep phase, after the sleep scoring experts had labeled the EEG data in our validation study. We used the data sets to train our algorithms, and it is important to know that we used what is called cross validation where the data set we tested was not a part of the training.
One challenge with sleep classification and sleep disorders is the first night effect. This is the effect of having all the PSG equipment mounted on the body, and the effect of sleeping in a different place then at home. Using Somnofy, this information is available with undisturbed sleep — our biggest advantage in comparison with PSG. The disadvantage Somnofy faces is that it lacks EEG information and a still missing medical accreditation. PSG also has a lot of specific sensors to classify sleep disorders which Somnofy does not have.
One things we do work on is detection of sleep apnea. Our preliminary studies show very promising results and we’re working with a private hospital to bring this on. With a trained eye you can see sleep apnea in today’s Somnofy as well.
JM: Sleep/fitness trackers have taken on new relevance in the era of Covid-19. How does Somnofy fit into this new view of consumer trackers as ‘predictive technologies’?
AB: There are two aspects of this I want to discuss. I use Somnofy on my parents and uncle to monitor their respiratory rate every night. With our 99% accuracy it is a fantastic tool to get early warnings if they are undergoing health changes. According to medical doctors the respiratory rate goes up in most cases of Covid and I can therefore look after my parents remotely.
And it is not only me 🙂 We have delivered many Somnofy sensors to elderly homes where they use that as an extra check for the patient’s health condition. Some municipalities have also installed Somnofy in their emergency rooms for Covid.
The other important aspect is that scientists now believe that one after effect of Covid is sleep problems and depression (among other things). To understand this effect and to find out how to help people is a problem we want to help to solve.