Sleep 1.6 Exploration of Novel Approaches and Analytics

• Finally,our exploration of novel approaches and analytics.I think I didn't spend any time sort of telling you a little bit of where Labfront came from. But we are a spin-off from the Center for Dynamical Biomarkers at Beth Israel Deaconess Medical Center and Harvard Medical School.
• And out of this lab, there came to a lot of new complex dynamical biological system analytical methodologies. Our main focus was in nonlinear dynamics and signal processing, and so some of the measures that came out from our lab includes detrended fluctuation analysis, multi-scale entropy, as well as cardiopulmonary coupling, which our lead PI, Dr.Peng. who is our scientific advisor, and Robert Thomas created.
• This is what you know that's used actually in sleep staging and many devices now as well. We've also had experience working, competing in the X Prize competition. This was back in 2012 and 2017, where we tried to, sort of create a device that weighed less than five pounds that could diagnose up to 13 conditions, and do it as well as the panel of doctors.
• We were a team of 350 teams and we were the runner- up winners for that. This a picture of us sort of winning, getting that prize in the process. Since we really have understood how to be clever with data, and for instance, for respiration, we found out that there are many ways that you could get to respiration using devices.
• So, for instance, for EKGs, you get what's called this EKG changes as you respirate because as you inhale, the lung impedance increases, and so the signals decrease, and you can see this change that could reveal sort of the inspiration expiration. There's also respiratory sinus arrhythmia that can also be used, sort of the spacing between each beat.
• PPGs can also, you could derive some respiratory rhythms from that by taking a look at the baseline of the PPG, and this has to do with the venous capacitance under the skin, which I could potentially talk about in the future webinars as well. And you could also use accelerometers. We found that if you have an EKG, EKG on your chest and you have an accelerometer on you on that we can actually derive respiration using the accelerometer on that as well. So it's really important to be clever about how you extract data from all the device data that you get from the various devices you have. We also think that the future lies in personalization and precision science.
• And for instance, since we do provide this capability, it took these settings. What is the best setting and can you personalize it and we can envision that you can collect data using raw accelerometer, using Garmin Labfront and a reference standard EEG, which we could hopefully bring into Labfront at some point.
• Take data for two to four days. Use machine learning AI to develop a personalized actigraphy approach where these settings are designed specifically for you for that device or for that participant. And then you could compare it against any standard actigraphy approach, which we know that, you know, for whatever they're out there, there's a single formula that's used for most for across the population.
• And then you could compare this against the reference standard where there's a PSG or EEG, do it for two to seven days, and then see if there's a difference in performance. You know, this is something that we would love to partner with people out there to sort of see if there's a way to sort of move this, the needle towards personalization and precision science.
• And finally, I think our vision is to make, as I mentioned before, a dynamic learning system for you, where you have this process of problem discovery, data processing, modeling, and then deploying the lessons you learn and then sort of creating a constant feedback loop. For instance, if you're trying to optimize sleep efficiency and you're trying to assess the impact of sleep on that, you want to make sure that you've got the stress and sleep efficiency correctly.
• You also want to do the analysis in an effective, in a rigorous way. And for instance, if you identify rumination as a factor where stress affects sleep efficiency, then you want to assess rumination, but maybe you incorporate EEGs, using beta or gamma powers and then so on. So our vision is to make this platform as dynamic as possible, but recognizing that in order for this to be possible, we really need to work on first the data integrity part of it.
• Also the availability of analytical methodologies and the openness and availability for everyone for that. And also the implementation in the real-world settings. So that you can really perform your research in a rigorous and scientific manner. So again, this is the way we see ourselves. We see ourselves not as necessarily experts of sleep or sleep research, but we do see ourselves as facilitators in your research so that we could help the patients and work with the device makers and manufacturers and app developers, as well as researchers to really advance sleep research. 

‍