Launched a little under four months ago, SleepScore Labs is proud to be the home of the newest technology disrupting the $60B consumer sleep industry. Developed by SleepScore Labs, a joint venture between ResMed, a 25-year leader in prescription sleep medical devices, Dr. Mehmet Oz, and Pegasus Capital Advisors, the S+ is a clinical grade, non-prescription, non-contact sleep monitoring device coupled with a mobile app. Involving over 10 years of development, SleepScore’s technology is the first step towards the company’s vision of revolutionizing the world of sleep science by setting a new standard for both clinical and consumer monitoring, analysis, and management of sleep. Medgadget had a chance to both test out the S+ technology and speak with Colin Lawlor, CEO of SleepScore Labs. Please find the interview below and the product review coming soon!
Mike Batista, Medgadget: Where did the S+ technology and SleepScore Labs come from?
Colin Lawlor: When it comes to sleep, ignorance is the greatest challenge for both clinicians and consumers. On the clinical side, sleep studies are the norm but they are not particularly accessible as they take a lot of time and effort. At the consumer level, it’s very hard to quantify sleep with self-reported sleep information being highly inaccurate and unreliable. It really is a public health issue as existing, subpar technologies provide consumers with misinformation about their sleep patterns and habits. We believe that if you can’t measure it, you can’t manage it. From that perspective, the team at ResMed set out to create a solution that quantified sleep in a meaningful way, in effect providing the calories of sleep, while lending itself to both clinical and consumer applications. The result is the S+ technology.
While ResMed does more research on sleep research than any other public company, we quickly realized that ResMed was not going to be able to do this alone and decided that aligning with the right partners would be the best approach to making our vision a reality. In January of this year, SleepScore Labs was formed as a joint venture between ResMed, Dr. Mehmet Oz, and Pegasus Capital Advisors, L.P. Dr. Oz is both an entrepreneur and a successful clinical personality who does an excellent job conveying the importance of sleep and sleep science. While ResMed is very strong in the area of sleep, Pegasus brings an extensive network of individuals and organizations who are not only experts on sleep but also experts on the relationships between sleep and other clinical areas such as behavioral issues and medication management. By complementing what each entity brings individually, ResMed, Dr. Oz, and Pegasus can collectively change the world of sleep at a larger level.
Medgadget: Let’s get into the S+ technology. How accurate is the device and how do you ensure accuracy when each person’s sleeping arrangement is slightly different?
Lawlor: The S+ technology is both a hardware and algorithm suite. The hardware is a movement measuring sensor that sweeps the subject 16 times per second and can see movement to under 1/10 of a millimeter. From that movement data, our algorithms assess the respiratory rate to an accuracy that is equivalent to polysomnography (PSG), the gold standard of respiratory monitoring today.
There are generally no issues with setup or user error affecting accuracy. The sensor sweeps in a broad, 30 degree view from the center of the device and includes a range gate of 4.5 feet meaning it cannot see any movement further than that distance. These specifications are based, in part, on the recognition that most individuals sleep with a partner so we needed to make sure we capture the target person’s data while not picking up artifacts from other people and things around the sleeping location. All that said, if the algorithms see irregularities either in the setup process or when the user is supposed to be asleep, it will recognize something is wrong and alert the user to reposition the setup.
Testing to find these specifications was done using labs with both robots and humans in various configurations to test all aspects of the technology. Probably one of the most interesting parts of testing was using a robot we call Brett. Brett is a breathing robot that can be used to mimic different human positions and, more importantly, different breathing patterns. We can set Brett up in a specific physical configuration with a certain respiratory pattern and then attach the S+ sensor to a separate robotic arm that moves to hundreds of locations around Brett. At each location, the system checks to see if we can still see his motion and if our algorithms are identifying the correct respiratory pattern.
Medgadget: How does the system differentiate between different types of sleep?
Lawlor: Our algorithm takes all the movement data collected by the sensor and breaks it up into body movement and respiratory patterns. Initially, a person begins in a stake of being awake during which they are moving a lot and have a high respiratory rate. As they begin to drift into light sleep, the respiratory rate slows down and the amount of body movement reduces. When the transition to deep sleep occurs, the body will still move but the respiratory rate becomes very slow and very stable. Finally, rapid eye movement (REM) sleep, during which people dream, is a state where the body is completely immobile, a precautionary measure taken by the body to avoid injury, and breathing patterns are highly variable due to the effects of dreaming. As you can see, all four states of sleep are characterized by distinct combinations of body movement and respiratory rate.
Accuracy in our algorithms comes from over 10 years and thousands of nights having the S+ in PSG studies all over the world. We have tested the system with individuals who experience “typical” sleep as well as those with sleeping disorders to provide the greatest range of insight into the algorithms we are creating. Now that the S+ technology is on the market, we have data from 2 million nights across more than 35,000 subjects providing a steady stream of insight to continually refine our algorithms.
Medgadget: Besides sleep data, what is the importance of the other information captured via the mobile app?
Lawlor: It’s not only relevant to capture body movement and respiratory rate data, which we use to determine the user’s sleep state, but also important lifestyle information from users such as which mattress they are using, if they are on sleeping pills and if so, which brand of pills, if they recently consumed any alcoholic or caffeinated drinks, or if they are stressed. With this information collected over such a wide population of users we are aggregating a huge dataset and have begun correlating back to a baseline on all these factors. This makes the dataset very powerful for deriving key insights about many interrelated factors that influence sleep. This in turn can be used to continually improve our feedback and recommendation engine that helps users improve their strategies for realizing a healthy, restful sleep experience.
Medgadget: There are other sleep monitoring technologies on the market such as wrist sensors or smart watches. How does the S+ compare?
Lawlor: In general, the weakness of other consumer sleep technologies is the quality and type of data being captured. Sleep technologies based on wrist worn accelerometers, for example, will never be good enough to identify an individual’s sleep state. Those technologies are very good at measuring limb movement for things like walking and running. However, when someone is laying in bed, movement is only one of the two pieces of important sleep information, respiratory rate being the other. Only knowing the individual’s movement was a good enough proxy for many years but in general it’s not enough data to assess sleep patterns.
Most people talk about sleep technology accuracy in terms of being able to identify sleep. But in general, that’s not a very useful measure because, logically, people who go to bed will end up asleep for most of the time so most technologies will be close to 90% right in identifying if someone is asleep. What really matters is the sensitivity to wakefulness. Currently, wrist sensors are about 20% accurate at identifying when someone is awake compared to the gold standard (PSG). More evidence is coming soon but we’re seeing 3-4x higher accuracy identifying when someone is awake compared to wrist sensors with the S+.
Medgadget: What is the company’s business model? As with many other digital health technologies, are there plans to provide an open API for developers?
Lawlor: Overall, our goal is to drive a better standard for sleep monitoring. To that end, we are currently not planning to open up the platform to everyone but are taking a very purposeful approach with how we make this technology, hardware and software, accessible. For the hardware, we know the cost of developing this technology is very high. This has historically presented a barrier to companies achieving higher standards. Therefore, we are making the technology available through licensing agreements with those existing brands who are committed to doing a good job improving people’s sleep. Similar to the hardware, the time, effort, and scale of data that has gone into refining our algorithms has made our software and analysis capabilities very strong. A second part of our business strategy has been to offer other companies access to our algorithms in order to analyze the data that they are collecting with their own devices and methods. Through the black box of our system’s intelligence, other researchers and businesses can qualify their data and assess the actual effectiveness of their own technologies in monitoring and analyzing sleep.
Link to purchase on Amazon…
Product page: S+ sleep sensor…
Link: SleepScore Labs…