Yeeeesssss!
He did mention above the interest in it:
Iâm just hoping the news from a bunch of people saying alpha 1=.75 doesnât equal LT1 (even though those studies donât really conclusively show that) doesnât mean recording this data gets dropped as even if .75 doesnât mean anything there is still use for the data. Also the discovery of the garmin app giving proper alpha 1 numbers using Ant+ should make it even better for TR as they should be able to easily do the same with their ant+ code. Not depending on BLE heart rate strap pairing makes the population of users that this could be useful for much larger and easier when you donât have to tell them how to pair.
Adding this functionality without users needing to know about it is why this could be very useful. It may be complex on the backend but no extra work for users.
Garmin and polar straps, thats probably a big number
Who cares? The TR app does the recording for indoor rides. For outdoor rides just post instructions. This isnât just about impacting the people whoâs HRV is being recorded as there may be other trends in data that this could make more clear and help train the model better to help even those without this data.
They may be trying to give the user the same data but that doesnât mean they are based on the same calculations. The details here are important. Garmin doesnât have a large data set of HRV info. By default no garmin even saves HRV data and if you do save it (hidden deep in the menu) and are paired with an ANT+ strap the data is poor (see discussions above on how the new garmin app gives good data with ant+) So if it records bad HRV data for alpha 1 calculation purposes (its very sensitive to small drops in data) why would you assume it can do those calculations in a useful way? Sure, it may still do some HRV measurements, but ones less sensitive to bad data.
Another thing to look at is that performance condition is from back in 2013. Alpha 1 is pretty cpu intensive for very low power devices like a watch from almost a decade ago so very good chance thatâs not what its based on.
Also while there may be machine learning to create the algorithms that run on the different garmin devices, the devices are stand alone and donât get their data processed by the backend. So their machine learning isnât learned off of your data and all calculations on your data are relatively simple
blah blah blah sounds like a bunch of red herring arguments.
Garminâs HRV/HR/power machine learning algorithms are working well for me, have been for about 2.5 years. Firstbeat has published papers with high-level algorithm details. I donât care how much TR points to its mountain of data, coulda shoulda woulda, why does TR devalue aerobic endurance in its training plans? After all, we are in the LT1 w/o blood testing thread. And TR isnât collecting HRV data yet, is still working on higher priorities, and while TR has stated that adding HR data is important for the future it is not here and none of us have any idea of when it might happen. A year ago they were going to support outside workouts, and a year later it morphed into waiting for Levels 2.0. Iâm sure that will be a nice improvement, but bringing new software to market isnât easy.
Iâm more interested in living in the real world and using whatâs available now. And I can tell you the Garmin stuff works, for me, no matter your attempts to cast FUD.
2 Likes
It often does but itâs complicated and contextual, like FB relationship status. Not sure itâs worth elaborating and the conversation has already moved on. LOL.
I can tell you this much, all this bitching about RPE is validating and vindicating for me. I struggle with it. I seek a metric or set of metrics, but I fear there is none.
In my previous life as a college runner, I always thought it was an asset to be able to lie to yourself. âBackstretch, you feel great budâŚletâs goâ <â thatâs always a lie.
3 Likes
All the garmin straps. It doesnât have to be BLE only if paired to TRâs app (which has been gone over a bunch of times above) That is already a large enough population size. As I already stated, sure the data may be the most helpful for users who have this data but as TR is based on building ML models, the extra correlations shown by this data in a large subset of the users will help lots of users
The TR app records straight off the strap. Why does the setup of a garmin head unit matter.
Garmin doesnât have the data from their large pool of users as this data doesnât exist. But somehow that is a red herring?
When did I say Garminâs stuff didnât work? I did just buy a Fenix 7 when it came out. All Iâm saying is we shouldnât be limited to whats available now as we should try to improve on it. Youâre the one who said that Performance condition from Garmin doesnât add value so say because that you donât think dfa1 would be useful. Youâre the one claiming garmin stuff doesnât work. Performance condition is not alpha 1 so what applies to one doesnât have to apply to the other. All Iâm saying is TR should look into this data
1 Like
These arenât such easy metrics to use. HR changes do to lots of other things that have little to do with stress like dehydration. The advantage alpha 1 gives is that it appears to correlate much better with physiological stress.
a garmin is a very low power cpu. TR runs on cell phones and full computers. TR can do more very easily and lots of people have the needed hardware
You know this, how?
AGAIN, look at what was said above in this thread. That is not true
I think this has been posted about lots of times so far:
Edit 2/23/22
Potential game changer - it appears this app takes advantage of a higher Ant+ sample rate than the Garmin native device. The result being - no more rate related missed beat artifact. I have confirmed this in my own recording and another from a friend. At this point we can have the best of both worlds - Ant+ to the alphaHRV app, bluetooth to the watch and Fatmaxxer.
Muscle Oxygen Training: AlphaHRV - the first native Garmin DFA a1 data field
Guess what, there is nothing special about this app. TR can easily read the Ant+ data the same as this app. TR does not need to read it like garmin devices and produce bad data. Stop assuming Ant+ data must be wrong
The H7 was considered good at the beginning of all the alpha 1 talk:
The H9/H10 may be better but that doesnât mean the older stuff wasnât good enough but researchers will try to just get the best data they can get. This is still new so more work on record the best data possible and analyze than in figuring out what devices produce good enough data. That comes later
Cycling and running are very different. The extra bouncing around and movement against a shirt (think swinging arms) running can cause more artifacts though some of that may be more strap related then electronics related.
1 Like
TR should look at HR and HRV, its been useful for me. The challenge is that things like HR zones are highly individualized (versus power zones below ftp). Didnât listen to the recent HRV podcast segment but did see the thread. I did a quick search and didnât see an obvious HRV feature request thread. You should start a feature request thread!
And regarding what I said⌠within a ride viewing of Performance Condition simply gives me an opening to be mentally weak if inclined. I look at it post-ride for big picture trends of fitness increases. Iâve been playing with HRV since HRV4Training app came out, first with iPhone camera, then HRM, then Apple Watch using breath app. Garmin 530 has been logging HRV per-ride for about 2 years, recently spent some time feeding rides into my MacBook Pro using the Collab Python script to output DFA a1. The short of it is that Iâve faith in using that Garmin feature as a trend/signal for potential slow fitness increases, and no faith in trying to use morning HRV as a readiness signal to influence what I do in the afternoon. But everyone is different, thatâs just me.
The Garmin ftp estimator uses hrv/hr/power and machine learning. My ftp is 270ish, did 32-min at 276 about 2 months ago. In the last 7 days I did a climbing ride with Garmin auto estimating 264 ftp, over-unders on Monday with Garmin auto estimating 260 ftp, and Wed night worlds doing a 14+ minute sub-threshold hero pull and Garmin auto estimating 267 ftp. Garminâs ftp estimate is reliably a bit on the low side, unless I go out and do a 50-70 minute long effort, but almost always within 10W. The Garmin auto FTP estimate is simply another signal that my slowly-getting-stale long test is still relevant (its a supplement to swagging estimates using interval power-to-HR).
Anyways, the point being that Garminâs use of hr/hrv/power and machine learning has been working well for me. Just like some of Apple iPhone on-device ML features. From where I sit it appears to do a good job at telling me when my Tuesday endurance ride was a little too high and is seen as tempo instead of low aerobic. But I canât prove that one.
Any idea how you get Alpha HRV working? It says to play with these setting but clicking on âEnter configâŚâ doesnât do anything.
You selected ble
Yes, but it doesnât matter. I cant enter anything into either.
My take on this is that any useful actual feature from TR using HRV is years out, and probably at least a year after they start collecting HRV data. If they have little to no HRV data for indoor workouts, this really hampers analysis and development of âMLâ based features. HRV may or may not be all that useful for TR, but they canât even start to answer that question without data.
This, along with interested users being able to get HRV data out of TR recorded files, is behind my suggestion that they start collecting this sooner rather than later. No UI changes required, little to no added support burden. If the started collecting the data tomorrow the soonest I would expect a feature based on HRV data would be mid 2023.
Regarding other things to do with HR, I expect using HR data for AT or redlight/greenlight stuff is a higher priority that improving analytics that just give users numbers.
1 Like
I donât know if it has something to do with the amount of sympathetic drive from being stressed, but, sometimes when my HRV is low I have the best workouts I ever have. RPE is super low despite a high power output. Its obviously not something sustainable but its an observation Iâve noticed about myself on occasion on quite a number of rides. HRV4Training Pro also indicates in the insight section âyour HRV doesnât appear lower after hard workouts, resting heart rate might be a better proxy to gauge recovery after hard daysâ which I thought was an interesting data point too.
1 Like
Two late nights in a row, travel last week, and a lot of work stress.
Oddly enough this morning I had my highest HRV in a couple weeks.
But I was tired and felt like absolute crap on the bike, Garminâs Performance Condition called it like I felt:
-6 is pretty ugly, donât often see it that low. End of ride my Garmin told me âYour training load is at a good level, but your fitness is decreasing. Your body may be struggling to recover, so be sure to pay attention to your overall health including stress, nutrition and rest.â
Yup. No more going to bed after 1am⌠Not a great condition to be in for my first early season kitchen sink ride. Endurance dropped to the floor around 2 hours. On the bright side I set season high 1-sec and 5-sec power records after 3 hours and 1700kJ 
Going to bed early.
1 Like
Iâd always thought that my Garmin just generated junk numbers for me but reading what youâve got to say about then I think Iâll give them another go. Doyou goth whole hog with sleep data from them as well or is it just workout related stuff?
1 Like
Just cycling workouts on the 530 (morning HRV on Apple Watch). Another ftp estimate at the end of ride - 265 - which was surprising given the grab bag of short efforts.
Others may not see the same results, itâs using HRV/HR/power, but this 530 and Garmin dual HRM is working well with my training. Nothing earth shattering, just reinforced I need to get more sleep after staying up too late a few times this week. Didnât need a computer to tell me that, not wanting to push out upper endurance after 2 hours was a good enough signal.
What does your ride look like?
Endurance plus short efforts, power in the pic above.