Sleep 5.3 Respiration & Temperature Monitoring with Wearables

We explore various methods of respiration monitoring using both laboratory devices and wearable technology like the Oura Ring.

Video Breakdown:

00:00 Introduction to Respiration and Temperature in Sleep Studies

00:11 Respiration Monitoring Techniques in PSG Labs

01:30 Wearable Devices for Respiration Tracking

06:25 Analyzing Respiration Data from Wearables

12:29 Temperature Monitoring in Sleep Studies

Now because, Oura ring incorporates respiration and temperature, I also wanted to go into that as well.
When you are in the PSG lab, the ways that you, acquire data about respiration, there's a few devices out there.
There are temperature sensors that sense the change in temperature, from inhaled in exhaled hair, air. They're typically, sort of placed near the nose or the mouth and are good for detecting apnea because it, it's, it's an all or no non sort of, evaluation of whether you're breathing or not.
Nasal pressures, transducers are also placed around the nose and they're sensing pressure changes that are associated with breathing. They're better at detecting hypopneas, or partial reduction airflow, because it's able to, it's not an all or none. It evaluates changes in the amount of flow, in the mouth or the nose.
And then you have these respiratory inductive plethysmographs, basically these belts, which could be piezoelectric and they're often placed on the chest or the abdomen. These are used, to assess respiratory effort rather than actual airflow. These are the devices that are used in the PSG lab to assess for respiration, but again, these are things that are not necessarily available when you're at home using wearables.
So what options are there to extract respiration? There's a number of them, that are available, when I'm talking, from the perspective of wearables. One is to use EKGs, and this is Movesense, which is, incorporated within our Labfront platform. And what you can't see here, so this is the actual EKG, and what you can do is take a look at the peaks of the R waves and you'll notice that there are some fluctuations here. And the reason that there is fluctuations here, is that it, and it reflects respiration and the reason you see that, and there are probably two reasons for that. One is that as you take a breath in, there's more air in the lungs, and as a result of that, there is, higher electrical impedance and the heart signal, particularly the QRS diminishes as you take a breath in.
So, you could see that, as you breathe in, you'll see a dip down here because the electrical signals, the depolarization of the conduction systems aren't seen on the chest, and so it dips down in the new exhale, it goes up. The other component may be that as you breathe, the axis of the heart changes as well, because as you lower the diaphragm, the heart lowers as well.
That changes the axis as well. So that also can probably contribute to these changes. And you can see here, the estimation of using this respiration derived from EKG is called EKG derived respiration or EDR. And this is represented here in this gray curve here. And it is very similar than the actual breathing, I think using a belt.
You can see that it does a pretty decent job of, estimating respiration, just using an EKG alone. Now, the challenge with this is that this does require the raw EKG data, and we don't often have that. I mean, we can do that, but it does take a lot of data. It consumes memory and battery life.
This is not something that is routinely available unless you know, you really focus on a particular time period and you just wanna do like a minute or two minutes or three minutes. You can extract the EKG and then derive the EKG derived that way. The other option for extracting respiration is using the PPG, and there's several ways that you could extract respiration from the PPG and, and, and there are probably four methods in which you could do this. So one of the ones that's probably more frequently used is the frequency modulation. It is the, it's the changes in the heart rate as you breathe in and out.
So if you breathe in, your heart rate increases. As you breathe out, your heart rate decreases. And so that's the respiratory sinus arrhythmia. And I know that that's what Oura uses to extract respiration. They are using, or at least based on what I've read of, their manuals and, and website, et cetera.
The others is to have, baseline waters, amplitude modulation and pulse width modulation. This relates to the fact that, there are some blood pressure changes as you breathe in and out. And there's several reasons why the blood pressure changes as you breathe in and out.
Now, when you inhale, you reduce, obviously, the thoracic pressures, and then you're causing more volume of blood to go into the lungs. And so there's a greater capacity there. As a result of that, less of the blood returns to the left heart, and the blood pressure, is reduced, because the stroke volume is reduced.
The other component is as you're breathing in more blood returns to the right heart, that causes the right ventricle to distend a little bit, and it causes the septum, the interventricular septum to deviate a little bit to the left. And so as a result of that, less blood also returns into this left ventricle again, causing the blood pressure to drop.
So as a result, as you breathe in, your blood pressure drops. As you breathe out, your blood pressure increases and that manifests is a change in the blood baseline wandering. Also a change to the amplitude, so you could see that the differences from the peak to the Nadir decreases as you inhale and then goes back up as you exhale.
And then the pulse width, the width of these also changes during the respiration. So there's a number of ways that you can extract respiration using PPG. But again, this does rely getting raw data from the PPG, which we don't have in our presently in our system, but we're looking ways to finding devices that can do that, et cetera.
So these are things that we're, constantly, exploring to, to provide this information for you guys.
This is actually, a demonstration of the data from our data science team itself. He wore basically a Movesense EKG also wore a number of, Garmin devices. And this is the RR interval. So this is the spacing or the beat, beat interval. The spacing between each beat. And this is when he's doing slow breathing in and out. When it's inhaling the, interval decreases, the heart rate increases in other words, 'cause the intervals an inverse of heart rate. And you could see that the EKG, which is the red one, you could see this, RR interval here, change. And then the blue, this is the PPG derived using the enhanced beat-to-beat interval from the Garmin device that we make available to you.
And you can see that it matches very nicely with EKG- derived RR interval. And then this is the raw signal that we used, using Biostrap. And you can see again the baseline wandering that I talked about, as you inhale and out. If you just take the high frequency component of this signal, you get basically do a high pass filter.
You have this signal, which again is showing the amplitude modulation as you breathe in and out. You could probably see, the pulse width, modulation as well. So you're seeing, again, the frequency modulation, the baseline wandering, the amplitude modulation that I talked about, that you could see on the PPG.
The other thing that I also wanted to show you that on the Movesense, you have a three axis accelerometer, and it turns out that accelerometer can do a pretty decent job of detecting inhalation and exhalation. So if you don't want to get the EKG or PPG, you could use, the accelerometer data from the Movesense to determine inhalation exhalation, and these are the three axis, of the accelerometer and you could kind of see these signals clearly. We could do a PCA to combine these and extract respiration from that. I'm sure you've heard of devices called, the Spire and I, I, I try to look in terms of how Spire is determining how, you know, the respiration intensity and rate, I am guessing that it's using accelerometer because it's not using any electrical connections, it's put on clothes.
So I presuming it's using this approach to get to respiration. The other, technology that's out there to detect breathing is Impedance Pneumograph. It's basically using electrical conductance or resistance or impedance to assess, for breathing. And it's the same idea that as you breathe in, there's gonna be more air.
Since air is not electrically conductive, it Increases electrical resistance and penance. And so these are ways that you could sort of another method to calculate, respiration, rate and potentially tidal volume. So not only how fast, but how big and deep the breaths are. These are the companies that are presently producing that.
Phillips has their Biosensor BX 100, Shimmer, Restmetrix Medical. To my knowledge, none of them provide us the raw data and I'm not sure if they're willing to share that data for the researchers in terms of integrating in a platform like ours. And also they're not that, affordable.
So I think, you know, that may be the limitation for us. So now that's kind of the way to assess respiration using those multiple methods. You have the EKG, PPG, you could use the accelerometer and then you could use Impedance Pneumograph. But then what is it that you're looking for in terms of the respiratory patterns to differentiate these different stages of sleep? So this is these plots of respiration, across time. I presume they used a belt to do this. And, in non-REM sleep, you, you should see a reduction, in respiratory rate, a reduction in the tidal volume, in other words, the amount or the depth of the breath, And then there's an increase in breathing regularity. With N two, you have, again, a more reduction respiratory rate, more reduction in tidal volume, more breathing regularity, and you can really see in N three here, a reduction in non-REM respiratory rate. You are breathing slower.
The tidal volume is reduced, so you see the amplitude is reduced. Again, there's significant regularity. This goes to sort of what I talked about, how, in deep sleep, particularly slow wave sleep, it's a time of consistency, regularity, and this is clearly demonstrating that. And for REM sleep, the respiratory rate increases.
However, the tidal volume doesn't increase, the depth is actually decreased, and the minute ventilation may be the least during REM sleep. And it's because, as I mentioned to you during REM sleep, there's muscle atonia. You're not gonna act out your dreams ideally, so, you know, there's an ingrained system to inhibit the muscles from activating, and this includes the thoracic muscles, the chest wall muscles. The only thing that's really, helping out with the breathing is the diaphragm. As a result of that, you're relying on the diaphragm. And this is why you can see what's called paradoxical breathing during REM sleep.
When you're inhaling the abdomen, sort of the chest sinks, the abdomen rises. It's because of the diaphragm is the only thing that's being activated. And so, you'll, you should see a reduction in tidal volume here. the breathing regularity again, as mentioned at, you know, during rem in general, there's just more variability and less regularity.
This is the summary of heart rate, heart rate variability, muscle activity, and respiration.
And finally, temperature. Basically, there's two areas that you could focus on temperature. There's core temperature and peripheral or distal temperature. For core temperature, during the day is during this white area and the shaded region is when a person's asleep. You could see that the core temperature sort of rises and it peaks in the late afternoon or early evening, then it starts to decrease before someone goes to sleep and it Nadirs somewhere during early morning hours.
This has strongly correlated with the circadian rhythm, and, it is believed that it prepares a person to go to sleep that's why there's a reduction in the core temperature before, someone actually goes to sleep. The challenge with collecting this data is that it often requires some invasive methods, either rectal probes, ingestible sensors, so it's not really ideal.
On the other hand, you, can get information from the peripheral or distal temperatures like the finger or the wrist. In this case, because it is exposed in the environment, you're gonna see much more fluctuation. And then the other thing to note is that it has the inverse relationship with the core temperature.
And the reason is that, in order for the core temperature to drop two things need to happen. You need to sort of reduce the internal heat production. You also need to increase the amount of heat loss. The way you do that is to dilate your vessels in your skin, your palm in your sole, and that causes your distal or peripheral temperatures to increase.
As a result of that, you see this increase even proceeding this core temperature drop, it's because this is the way that core temperatures is being reduced. So as a result, you see this inverse relationship that you have a peak of the distal temperature happening in the middle of the night as opposed to the Nadir in core temperature. Nadir is during midday, and this is regulated by circadian controlled vasodilation and constriction. At evening you have vasodilation. This increases the distal skin temperature, enhances heat loss and leads to the core body temperature decrease. As a result, there's a little bit of a phase advance.
You see this peak happening before the Nadir of the core temperature here. Now daytime's opposite your vasal constriction of the skin vessels, which leads to decrease in the distal skin temperature, and this leads to conservation heat, and then increase in the core body temperature.
So, this is the summary of what we see in the core temperature. We see a decrease, over time, of the core temperature increase in, distal temperature. In REM sleep however, there is what's called poikilothermia. There is loss of sleep regulation and so you don't really see any type of patterns that basically you should see your, these temperatures go towards the environmental temperature, so if you're in a very cold environment, your body temperature will match that.
The reason I sort of use this, thing is that the distinction in these, each of these stages are not as clear because this is more of a circadian the whole day and night cycle. This is why I created this graph here.

Course episodes

Sleep 5.1 How Consumer Wearables Measure Sleep Quality

Sleep 5.2 Heart Rate, HRV, and Muscle Activity During Sleep

Sleep 5.3 Respiration & Temperature Monitoring with Wearables

Sleep 5.4 Challenges in Using Wearable Devices for Accurate Sleep Monitoring