In my case, the results are completely wrong. During this workout I had a respiratory rate of around 17-20 and not 30-35 as your app indicates. The garmin estimate is already closer but still overestimated. Polar H10 heart rate monitor. Anyway, the AlphaHRV in general miscalculates my oxygen threshold. According to AlphaHRV I have an oxygen threshold of about 230W, and this is closer to my LT2 than LT1 
Translated with DeepL Translate: The world's most accurate translator (free version)
The measurements are fully validated by Bruce Rogers’ team, Garmin grossly underestimates the measurement.
I don’t know how you have estimated the aerobic threshold with the respiratory rate but it is a data that has not come out yet, we use a software that is being created to do it, I doubt that you can do it manually. The aerobic threshold is calculated with heartrate and then we associate some watts in the change of tendency, all this we have with several lactate tests.
You also have a lot of artifacts (not normal), you should look with Fatmaxxer if there is a deviation in the respiratory axis and you have to move the heart rate monitor somewhere.
I didn’t estimate any threshold, I counted my breaths for a minute and came up with 17, AlphaHRV showed 33 and the garmin showed 22.
Alpha DFA1 0.75 I have at 230-240w and my FTP is something around 240w.
have you used a controlled ramp ?
I think we are talking about totally different things, the respiratory rate we calculate has nothing to do with counting breaths, we calculate the sinus respiratory rate, which is a different concept and is calculated through rr intervals.
Thresholds are not calculated through Dfa-alpha-1 and even less with standard values of 0.75 and 0.5, just as nobody calculates lactate thresholds in 2 mmol and 4 mmol, the sinus respiratory rate has a high correlation with the behaviour of lactate, so we can detect two inflection points or two trend changes, through multivariable calculations we can detect the first threshold, assign a heartrate value, watts and a customised Dfa-alpha-1 value.
Breath-counting is not validation of this data.
We will start a validation with more than 1000 laboratory tests.
It is a test very similar to the lactate test, we do 15 minutes very soft (40% FTP) to “clean” lactate, then we do ramps of 6 minutes with an increase of 10% compared to the first ramp.
40%-50%-60%…
Did you take the test (according to the protocol) or was it normal activity/training?
regards
Artur
I take de test according to the protocol
My friend @Luisma_Gallego_Soy_P, I asked my colleague @Wojciech_Marszalkows if he was performing the test according to the protocol, because if not, it was pointless to say that the method was wrong.
regards
Artur
During the test (plus 20w every three minutes) Alpha DFA1 0.75 I have at a power closer to LT2 than LT1.
And with this breathing I don’t understand anything anymore I count myself as having 17 breaths per minute and AlphaHRV tells me that I have 35 breaths per minute. But @Luisma_Gallego_Soy_P writes that this is how it should be because these are some different breaths
The algorithm might detect a slope in HRV, pos or negative, and count slopes. Then 35 is double of an entire breath cycling 
and thus more precise then counting full cycles.
Be aware though, that you have to much artefacts! I had to replace the strap on my H10 after a little more then a year to get rid of artefacts. Have been using the strap from my Garmin Dual since and that works very well. The straps are interchangeable.
When testing, the Firstbeat (Garmin), Movesense Medical and alphaHRV algorithms were used. Garmin underestimated the measurement, alphaHRV was very close to Movesense, the version with respiratory frequency was distributed for those tests that were done, and when it was seen that they were practically the same, it was decided to distribute the new version.
Pls, up to?
tks
If you’re only interested in AeT, 90% should be enough. Only one deflection point would be found. If you want both AeT and AT, I guess 130% should be more then enough to get both deflection points.
tks,
had no idea this could “estimate” AT also.
But I think that above 100% would be hard with 6m intervals.
Any tips?
Regards,
If you want to find both AeT and AT, go to failure as you do with a ramp test.
You don’t have to come back down with this method. And it’s not a1 that defines the thresholds, it’s the breathing frequency deflection points who define the thresholds and then those translate into a a1, power value and HR value to define them.
The research on this shows that breathing frequency tracks almost identical as lactate.
Read up before blindly using something.
I have.
But each one has “his own protocol”…
And it has been suggested 6m intervals with 10% increments.
Nonetheless I’m most interested in alternatives of AeT other than lactate.
If you wanna comment and/or help great but your last comment is totally unnecessary.
The most accurate way of defining both thresholds (apart from an expensive met-cart test) is lactate measurement and, importantly, not using fixed levels of lactate but detecting deflection points in the lactate curve during a slow ramp protocol. The first deflection point from a stable lactate level to a slowly rising lactate level is the AeT. When the slope of the rising lactate level increases, there’s a second deflection point, defining the AT.
Early stage research now might show that breathing frequency is highly correlated to lactate, and there is maybe a good chance that the deflection points can be evaluated from breathing frequency. If this proves solid, we have a way of more accurately measuring both thresholds without having to rely on lactate measuring devices.
A reliable way to measure breathing frequency is an absolute necessity. And until now, you needed a respiration measuring device to do it accurately. That’s not a big advantage compared to direct lactate. Garmin’s Firstbeat algorithm for resp rate (derived from HRV) was found unreliable in several comparing studies over already a number of years. But the guys from AlphaHRV seem to have found a more reliable algorithm that is very close to the Movesense device.
That’s where it stands as of today.
An oversimplified view of what the data will show is this:
Where section 1 is the part below AeT. Lactate and apparently resp rate is as good as stable even with rising power in this low intensity section.
In section 2 lactate rises in a linear fashion with HR/Power. If you would stop raising Power, lactate will stabilize too, maybe even go down slightly. Your body reuses the produced lactate as fuel and is able to keep up.
When you keep increasing intensity, you end up in section 3 where your body can no longer keep up clearing lactate and it goes through the roof. Your legs explode…
The better your aerobic condition, the more those points shift to the right.
If breathing rate shows a similar behavior, it is a question of developing a software tool that is able to detect the three trendlines and the cross points. This requires highly skilled mathematics and knowledge of exotic filter mechanics to ignore outliers and artefacts. Until that tool is available, it makes no sense to discuss protocols.
That test is wrong, you can’t create a cool down, if you want AeT do up to 90%, what is the problem you are going to find that you need a software that sees that trend change as accurately as possible, that is exactly what we are creating.
Indeed, the Firstbeat algorithm has little validity in the measurements, thanks to the collaboration of Bruce Rogers, we have been adjusting the algorithm, we have to think that we are fighting against the low processing capacity of the Garmin equipment, even so we have achieved something that can be worthwhile. We are currently developing an app that detects thresholds, Bruce Rogers commented on the high similarity of the respiratory rate with the behaviour of lactate and we have taken advantage of this to see the same behaviour in the respiratory rate, soon we are going to do massive tests, all with laboratory tests. We are financing all this with our own time and money, so we are going somewhat slowly due to lack of resources.
