Testing for Aerobic Decoupling

LT1 is the top of Seiler’s Z1. I’ve used that HRV app this week and using the suggested protocol it put my LT1 at 78% of my max HR. I’ve been doing my easy at 67% of max. I do wonder about working closer to this 78% figure. I’ll do it again next week to see if the protocol delivers consistent results for me.


Reporting back on the use of HRV4Training’s HRV Logger and their DFA analysis to identify the aerobic threshold. Last night I did a slow aerobic 8k run, and the whole run was apparently above the .75 threshold of DFA as indicated by their research. 0.75 is the threshold they believe indicated the aerobic threshold.
This morning I did Petit with the logger running and the live display. Attached is the screen shot from the end of the ride. It does appear to show that as the ramps increased I did get closer to the aerobic threshold. This matches with both the heart rate data from the ride and power data from the ride - both which I have reference values from laboratory testing 9 months ago.
There was one outlier which is that single reading at 0.85, assume that is the sort of data foible that can occur; although it did occur at the second ramp when I was really only starting to warm up; but it was gone 2 minutes later.
Tomorrow I have threshold intervals and VO2 max intervals. Let’s see how it works then…but so far so good.


second report on the HRV Logger. It appears not to really work on short intervals like VO2Max. Probably not a surprise, the intervals were 2 minutes and the sampling period is set at two minutes. It did show increased workload and getting closer to aerobic threshold but didn’t register any of the intervals as over threshold.

Today was a short progression run.

I’ve attached the aerobic/anaerobic contribution chart from WKO along with the HRV Logger screen. It does align fairly well and of course it is real-time where as WKO is after the fact.

I’m pretty pleased with it so far and for steady state endurance rides when you want to ensure you are staying aerobic this seems close to perfect…

@Nate_Pearson or @ambermalika this almost fits into the branding of ‘Adaptive Training’… Any chance you could calculate DFA (alpha 1) in TrainerRoad and use it to control intensity of long endurance rides??


Looks like the HRV Logger app is now out for Android users: https://play.google.com/store/apps/details?id=com.asma.hrvlogger


I messed around with the HRV logger app for android this morning a bit, without fully reading the instructions :slight_smile: Initially it looks like my LT1 may be lower power than I thought, but I need to do a few more experiments. My HR was higher than ‘normal’ for the power I was doing today, not sure why it was different.
From reading the paper this is based on it sounds like 0.75 is likely a rough approximation, and more work will need to be done to firm up a cutoff, but that it is definitely higher than 0.5. Sounds like it is still early days for this, but all very interesting.

What’s the conclusion when decoupling is negative?

Don’t think there is a conclusion.

I’ve seen negative values on rides where I struggle early on and don’t warm up properly. Heart rate goes through the roof and then settles down. I usually see that with larger percentages. A negative value for a fraction of a percentage might be a good thing. Or maybe not. I’ll consider how I felt throughout the ride if I get an odd AD value.

It was a fairly normal endurance ride, steady Z2. I felt good. Perhaps what I can say is that 2h is not enough observe decoupling at this intensity.

Basically anything below 4% (including negatives) just means you haven’t pushed enough to find your breaking point - so you need to increase intensity or duration to find where you start to decouple.

Two hours is the minimum, but people with a strong base need more time to find the actual break point. Just as a reference, below are some of my endurance ride durations and decoupling from the last two weeks

3:15 0.8%
3:05 -0.4%
3:45 2%
3:00 0.1%
2:00 -4.3%

1 Like

Agree, but I would add, it’s a good maintenance session. You didn’t waste your time.

My Wednesday 2hr Z2 always looks like that. I push it on the weekend rides.

1 Like

Yeah I didn’t mean to imply it was a wasted session, you don’t have to push to your breaking point every ride…a

1 Like

I hesitate to bring this up but I’ve been exploring IN_EX score lately. It is more sensitive (and seemingly more useful) than measuring decoupling with Pw:HR first half last half, which is quite crude.

It’s an attempt by Seiler to quantify stress / strain. I’m still getting wonky graphs occasionally so I’m not completely fluent yet.

:+1: Oh yeah, and not directed at you (even though I quoted…my bad). I know you know your stuff. Some people might come across that and go “shouldn’t I be getting a ‘high score’ if I want improvement?” Well, yeah. Sometimes :grinning:

I went out today for another Z2 ride, this time 3 hours and outside. I pace these rides by heart rate. Had to battle some crazy wind, ended up with same HR, more watts and 5.5% decoupling.

I got the HRV logger for Android, and have used it on about 6 or so rides, all steady state efforts. I’m using a tickr2 as my HRM, and I don’t think I’m getting good data. I’ve attached 3 screenshots from my latest ride, and the number of artifacts removed concerns me, as well as that I get a blank screen on the ‘alpha 1’ metric when I hide noisy data. Looks like the data is very noisy, which would explain how variable the data is when I am just riding steady. The spikes in the middle and end are when I was off the bike. I might spring for one of the recommended Polar HRMs to see if that fixes this.
@marco_alt Is this the kind of problem you have seen with Tickrs?

1 Like

I’m using Tickr2 on iOS and not having any issues with artifacts. Only spikes are when I get off the bike

Are you paired over Bluetooth or Ant?

1 Like

hi @toyman, thank you for trying this method and the app.

Yes, this data is certainly unusable for HRV analysis (DFA or other) due to the very large number of artifacts detected. You might be able to see some of these issues for a shorter ride also by looking at the RR intervals time series, there might be more “high-frequency noise” on the signal, which should otherwise be quite flat due to the naturally low variability during exercise. Another check could be to look at rMSSD for the same recording, you would expect very low values 2-5ms, while if there is more noise, it could be a bit higher.

Unfortunately, the Tickr has provided inconsistent results so far, even here I see just the message below yours @stevemz is reporting data that seems accurate (it is not really possible to tell without looking at the actual waveform, so we can only try to estimate the quality by looking at abnormal beat to beat deviations, which is a decent method).

On Android we have not enabled ANT for this app, so the protocol does not seem to be the problem (which instead might be the problem when using .fit files as previously reported also on this forum, when we didn’t know that ANT was very prone to packet loss).

If you have the opportunity to try a different strap or are interested in changing strap, then my recommendation is always the Polar H10, which works with both Bluetooth and ANT and therefore can be also used with other tools

1 Like

@toyman definitely recommend the H10, it’s working well for me, even with running. I have had the odd recording with artifacts but it is very uncommon; I think there must be the occasional time when I don’t sit the strap correctly or something like that.


Thanks Marco - H10 is ordered.

1 Like

I’m missing something about how to test this. Just read the TP guide and they talk about first half and second half of the ride. Then calculating the ratio differences between them.

What if I ride for 2hours, and then the decoupling only happened 10 mins before I got home?

How would you work it out. Something wrong with how I’m understanding this, something simple I’m not getting.