Sleep 7.5 Traditional vs. Modern CPC Methodology

Unpack the methodology behind CPC analysis, from traditional coherence-based approaches to new methods emerging today.

Video Breakdown:

00:00 Traditional CPC Methods

01:19 Calculating Coherence & CPC

02:20 New Methodologies & Advantages

03:36 PPG-Based Respiration Extraction

04:15 Future Applications

And then the the next one, I just quickly wanted to go with the analytical methodologies and I don't want to go into too much of the mathematics 'cause it can get a little complicated here.
But what is done in the CPC methodology, this is the traditional one that was done back in 2005, is you're taking the RRI data from either EKG or PPG, and then you're doing the respiration data, again, derived from the EKG, and then you're taking a window of about 1024 samples at two hertz sampling frequency rate, which equates to about 8.2 minutes.
And what you do is you divide this window into three sub windows, each about 512 samples long, or 4.1 minutes. And to each of the sub windows, you calculate the Fourier-transformed to both the RRI data and the respiration data. You calculate the cross spectral power, so this calculates the amplitude for each of the frequency components, and then that also creates the phase difference between the respiratory and cardiac components.
And then you apply these three measures and apply it to the full window and to calculate the coherence. So what that means is that, so let's just say the R uh, the heart rate and respiration has a lag of, you know, let's say 180 degrees, in one window. You wanted to see whether this lag in phase is the same for across the three windows.
If it is the same, this coherence value becomes one. If it's not the same, like if it's a hundred eighty, ninety zero, then it becomes close to zero. And then you calculate the CPC. So this one calculates the phase coherence, this is the amplitude, and then you get the CPC, and then you shift the window by, 2.1 minutes and then you repeat it.
And that's how you get the values that you get on the CPC reports. What are the limitations to this traditional CPC method? Now, this CPC method, ever since the 2005 has basically not changed, it's been the same method and it's ones that's repeatedly been used in subsequent publications.
And it's interesting that it sort of relies on the Fourier-transform. There are many new methodologies that are coming out lately. I mean, they've been around for a while, but the Fourier- transform here utilizes a large window of about eight minutes. So that prohibits more granular analysis.
This might explain why it may not correlate that well with the staging that you get from the sleep epochs, because the sleep epochs is about 30 seconds. Fourier-transform also doesn't do terribly well with non-sinusoidal waveforms. For example, the Phase A episode during CAP happens at, an instantaneous time.
Obviously it lasts about, you know, 15 seconds or so, but then you have a lag period about that. And so that waveform is, is not necessarily sinusoidal. And so the Fourier- transform doesn't necessarily do all that well for that. The phase coherence that I talked about compares the phase differences across three sub windows.
So again, that's just three sub windows, not that great in granularity. The CPC also doesn't do necessarily great job at distinguishing between REM versus wake, and it's not applicable with patients with arrhythmias. Those with excessive ectopic beats, AFib or paced rhythms. The other challenge is that the traditional method use EKG, but how do you extract respiration using a watch, which uses PPG?
And there's probably various methods for that. One of the things is that during, PPG, you have respiratory modulation of that PPG, there's a baseline wandering. There's the amplitude modulation, so the, the difference between the peak and the Nadir, which also changes, and also the pulse width between the two.
So there are a number of ways that you can extract respiration using PPG, but it's not something that necessarily has been studied broadly. I would guess that that's what's being used by SleepImage because this is the sort of figure that they have on the website, which kind of indicates that that's how they extract respiration.

Course episodes

Sleep 7.1 Understanding Cardiopulmonary Coupling in Sleep Studies

Sleep 7.2 Cardiopulmonary Coupling: Sleep Staging Classifications

Sleep 7.3 High vs. Low Frequency Coupling in Cardiopulmonary Physiology

Sleep 7.4 Cardiopulmonary Coupling and Sleep Apnea Detection

Sleep 7.5 Traditional vs. Modern CPC Methodology

Sleep 7.6 Cardiopulmonary Coupling in Sleep: Strengths, Limitations & Future