You seam to have a good experience with HRV’s and you mentioned Garmin.
I wonder you could have the answer for some of my doubts:
I’ve given up of daily measures but I’m really interested on checking my LT1 from time to time.
Due to several nuances of this I only look that when following the SBC - AeT Test DFA-a1 protocol always taking care of trying not to move during the test etc.
Do you measure DFA-a1 on several kinds of rides or you follow some protocol?
I’ve heard that BT is better than ANT+ but others say this is not true. Fact is I can not add my Viiiia to my Garmin Fenix 5x. it simply only shows the ANT+ id (works fine with others devices…) Have you have the chance to compare (or are aware) a session with BT and with ANT+ to confirm one is really better than other?
Appreciate any help.
I did not check ANT+ vs BLE myself but Bruce from muscleoxygentraining did. Since reading his remarks I only use BLE for my HR sensor.
I do have a comparison between Garmin Dual and Polar H10 since I bought a H10 recently. I can assure you that there are a lot less artifacts caused by missed beats on the Polar.
About your HR sensor: is there a setting allowing the sensor to connect to two different devices in BLE? The Polar has this setting in the Polar app but the sensor comes default with that setting disabled. I had to enable it, to run Fatmaxxer on my phone and simple HR on my Garmin Edge simultaneously. I could have connected it in ANT+ to my Edge but preffered BLE because the Garmin records RR intervals too and I do analyse that stream post-workout.
Another thing: I had to pair it first ANT+, disable (not delete!) that sensor, and then search again before it found the BLE stream. Might be worth a try.
I do measure dfa-a1 on all my rides now, but I pay most attention to the live value during endurance rides. I did one “religious” endurance ride to check my AeT HR and will follow up every couple of weeks.
I’m not sure what you mean by the AeT test but from the picture it looks like you ramp up slowly untill you get under 0.75 and then go down again. I’ve never done that kind of test. I just ride endurance with a1 around 0.75/0.8 and then check my average HR and Power for the zones where a1 is stable. I do see now on regular rides that when I keep my Power at that value for a period of time, that a1 stabilizes around 0.75-0.77. That’s enough to know that I’m riding Seiler’s S1.
Regarding daily measuremants, I have very good experiences. Just last Thursday, when Belgium won his football game, and I went for way too many drinks afterwards, my daily HRV value on friday morning reflected that bad choice… Just as it did when I did my first 20k run three weeks ago. But you have to be very consistent and do the measurement in a relaxed state first thing in the morning.
Re Python vs Java for this: Intervals.icu is written in Java so its easy to add new calculations in Java (and its fast). I could call a Python implementation running as a separate service easily enough but Java is less dev work.
Ahaaaa, I didn´t notice that Notification option for now (Work and life doesn´t allow me any training this week
But the audio notification sounds like a great option! I will definitely try it out, thanks for the tip…
Besides, I heard on a podcast about one interesting study that was made on competitive amateur cyclists (not Pro Tour, but experienced avid racers) and they gave 2 options for their Garmin units
nothing but remaining distance
whole bunch of metrics (as much as display could take)
And they let them do 30km time trial…
Your suggestions?
Mine would be 2) wins because they can pace better…
Well, apparently because of the cognitive overload (they did a good job to eliminate other factors), subjects made in average only about 220W, but with just the distance, they did in avg 280W…
Sounds unbelievable, but apprently our mind can´t separate physical fatigue and mental, so when you are overloaded with metrics, you can be drained much more quickly… So maybe a voice guide would be a very good option maybe the best would be a nice soft voice telling you: “good boy, that is perfect / good tempo, you can push sliiiightly more if you wish” and a bit firmer voice: “easy boy / ou ou, this is waaaay too much, chill!”
After all, voice guidance is THE thing for the human pickers in warehouses, it seems it is the best way how to guide human brain to “just follow”
Calculation of dfa-a1 should be done over a 2 minute window, so I don’t understand the 2%/min. There will not be one single 2 minute window with a steady state.
The double sloped ramp crosses the Aerobic treshold and not the Anaerobic.
We could tweak this to reflect the 2 minute window but have never heard of it before.
What you think?
Would we change all intervals to 2 minutes or just 2 ou 3 right before the peak?
I’m just looking for a protocol whatever it to be.
Bruce from muscleoxygentraining and Marco from HRV4Training are the experts on this thing.
HRVLogger and the Colab implementation divide your session in 2min windows to calculate DFA-a1. It is a “detrending” calculation, meaning you need enough data to detrend. And a1 changes with effort, thus you need a constant effort during that calculation window.
With fixed 2min windows, you have to discard the first (effort change - no steady state), calculate the second and confirm with a third. That’s 6 min at the same intensity to have a meaningfull calculation of 2 DFA values. And windows with too many artefacts have to be discarded …
Fatmaxxer app is not using consecutive windows but has a rolling 2min window, recalculating every 10sec (configurable). So here, you have 7 calculations in 3min at steady state.
Marco’s recommendation on the Colab sheet:
Hence, my recommendation would be to use either a workout with a stable / constant effort at low intensity (and maybe try a few files to see if you can find differences in alpha 1 just below or above your aerobic threshold), or to use a progression workout, in which the intensity is relatively stable for at least 5-8 minutes at each step. This way we should get features representative of what we are trying to capture, otherwise the effect of recoveries, short intervals, etc. - will create issues for this analysis as we are not looking at instantaneous heart rate values, but we are using a few minutes of data to compute HRV features
I’ll read it.
5-8min. very different from the other.
I’m using Garmin as logger and Golden Cheetah to analyze it.
Ale Martinez has update the chart adding AeT pace/power estimates with the same algorithm for AeT heart rate, that is linear regression using averages from 2 min segments looking like this:
I hadn’t seen the Golden Cheetah implementation yet. Will check that out and try to compare different methods.
Do I need a special version of Golden Cheetah or can I import the chart/calculations?
I have changed how HRV data is stored in Intervals.icu. It now associates each tuple with a timestamp so it maps back to the power and other data (tx @John_Peters_endura.f for the tip). Unfortunately you need to reprocess old files to get this.
This is what the CSV download looks like now. The last column is the HRV data with a ‘:’ between each beat to beat ms value:
On another note I noticed that my Garmin Edge 130+ stopped recording HRV data recently. I suspect letting it update undoes it. Have to repeat this process: HRV-Guided Training - #19 by david
I’m using a H9 and I believe the only difference to the H10 is that the 10 has dual band recording. FatMaxxer app definitely works with the H9 and I’ve sent files to Bruce Rogers and he has been able to use Kubios to analyse them.
Ian from Fatmaxxer would appreciate if you gave him the prove that H9 works just as good as the H10. He only tested H10 so he’s not going to state that H9 is working without being sure.
The first lines in the ReadMe contain
This app requires a Polar H10 (or possibly H9).
Just head over to the Fatmaxxer Github page and give him the good news.