I’ve been approached by a number of companies seeking my perspective on the viability of their sleep health product. Unfortunately, however, nearly all of them are hashing out some iteration of the same flawed technologies that already plague clinical sleep medicine. Alternatively, they have such grand yet scientifically-unfounded ideas that what they're hoping to do (and at the price point they're hoping to do it) is completely impractical...which is saying a lot because I'm willing to dream big.
Knit stood above all the other sleep technology out there because its tech offers the potential for looking at sleep in a way that no one really has before: by providing a continuous, personalized baseline of monitoring that surfaces meaningful changes and abnormalities that could help evolve our understanding of sleep and sleep disorders and improve sleep health for everyone.
How fortuitous that a friend had asked me to speak with David Janssens, CEO and Co-founder of Knit, as a favor. I was surprised by the practicality of Knit's product, while also impressed with the capability of the tech, which was twofold:
- Knit has state-of-the-art machine learning that does what no human could do
- There is still the possibility to verify and explore visually the peculiarities that can't be accounted for by the untrainable aspects of an algorithm, aka the infinite variability of human disease.
My sentiment for why these two facets are massively important to the future of sleep medicine has been best expressed by the Stanford Big Data in Biomedicine conference: Machine learning/AI won't replace doctors, doctors who use machine learning/AI will replace doctors who don't.
Knit can bring sleep science out of the dark ages
Knit’s tech provides a potential to fully understand sleep, far more than the algorithm that runs without updating and adapting, which is typical in the field of sleep.
Knit takes big data and uses AI to learn, adapt and enact findings quickly in the field of sleep science
Sleep studies try to distill big data into a few inaccurate numbers, highlighting how little science is willing to explore and apply our vast research knowledge. There is also the common problem in medicine of a 10-17 year gap from discovery in research to clinical adoption. There are too many findings that have been repeatedly verified, yet have never been implemented in doctors’ clinical practice.
Clinicians also continue to choose to apply demarcations for sleep scoring that are generally arbitrary or were created with technological limitations in the "era of ignorance”, meaning that even though the science was sound in early research, it was inherently flawed based on incorrect or incomplete knowledge and capability.
Knit has the ability to surmount the obstacles of traditional sleep science limitations by using machine learning to analyze big data to create accurate scoring in a tighter gap of time between research and clinical adoption. The benefit of Knit’s technology goes beyond the unparalleled camera data capture capabilities by expediting the process of aggregation and conclusion with the AI engine to adapt and quickly evolve our understanding of sleep in general.
Knit looks at sleep over time, both individually and broadly, to create an evolving, holistic baseline of sleep
For the most part, right now sleep is clinically assessed based on a single (or a very few) disturbed nights of sleep that don't reflect the reality of the human experience. Even in day-to-day life outside of assessment, we don't watch each others’ sleep on a regular basis. These factors combined have resulted in clinical knowledge of sleep that has a sampling bias for people with sleep problems based on too limited data.
An expanded view that takes into account the individual and daily fluctuations of sleep is a huge benefit of Knit’s tech. Looking at the whole spectrum of human sleep will be truly something interesting and could finally lead to a complete understanding of how sleep works - for each individual and for the general population. For individuals, Knit can enable the person to watch the natural evolution of their sleep, while also collecting an ongoing baseline. For me as a clinician, this baseline can be incredibly valuable, as I can have a perspective on what is possible to achieve if someone comes to me with a sleep issue. Most people come to me only when a problem has developed, and usually after it's been there for a long time. This thwarts me from effectively counseling them on what results they can expect to achieve with treatment.
Ongoing sleep monitoring can help individuals detect and address issues
In addition to the baseline that Knit can provide, the AI technology has the amazing power to detect deviations or issues early on, either based on the individual’s baseline data or on standards calculated with the data from Knit’s user population. Early detection can result in a simple course correction that can prevent people from developing a sleep-wake problem that affects their whole lives.
Another value that Knit provides to me as a clinician is meaningful pre-/post- assessments of therapeutic interventions. Much of sleep medicine relies on patient reports, with the occasional sleep study or 2 weeks of actigraphy thrown in. And from what we know about human biases, self-reported data is not particularly useful. For example, if I'm talking to a CPAP patient, I will ask how the last 3 months have gone. Even if the preceding 89 nights were good, which data could verify if it was available, if the night immediately before was bad, then the patient would report the whole 3 months as "awful".
Humans aren't generally obsessing over the details of their sleep, AND they often don't observe their sleep. But if there’s an ongoing measure of comparison, based on an aggregation of accurate data, then we have something to base our decisions on. Even when a person’s sleep is OK on average, there are still random "what happened this night" aberrations. If the answer to that question could be found, then each night of sleep could be great, and people could be living their best lives.