DFA Alpha 1 (DFA a1) – new threshold discovery method with HRV
Marco Altini is probably the world’s leading expert on HRV & HR usage in sports & fitness physiology and he advises several of the leading sports companies, for example, defining RELATIVE EFFORT for Strava which I talked to him and wrote about (here) a while back.
In this article, Altini talks about a new method of determining LT1/AeT during exercise using Detrended Fluctuation Analysis (DFA Alpha 1). You might hear a lot more about this over the coming years.
To Be Clear – Aerobic Threshold
This is NOT the LTHR (LT2, AnT, OBLA) that athletes normally talk about.
- ANaerobic Threshold: LT2 is the point above which you accumulate lactate problems (Z4/Z5 boundary)
- Aerobic Threshold: LT1 is the lower point which marks a shift in your energy sources from fat and more markedly toward CARBS. It’s the point below which you can still easily hold a conversation whilst working out. (Z2/Z3 boundary). This is the one we are talking about.
However, both are really important for endurance athletes and LT1 is also super-important for significant parts of the population who are using exercise as a route to weight loss and general fitness.
Indeed these thresholds are SO important as they actually represent REAL ‘zones’ that exist in your body. Broadly, the other zones that are defined in 5-zone or 7-zone models don’t really exist but are useful as a means to place your efforts on a somewhat complex spectrum of physiological responses.
Defining Workout Zones
Regular readers here have almost certainly defined training zones based on either LT2/LTHR or HRmax. Hopefully, all of you realise that the formulae for HRmax are somewhat imprecise and that a proper LTHR test is actually quite demanding. Furthermore, your tests and training will potentially be significantly impacted by things like fatigue and caffeine. No doubt, you take both of those in equal measure 😉
Well if your LTHR/HRmax is wrong then there’s a very good chance that your zone formula comes up with an incorrect value for the lower LT1, Thus your aerobic training could be wrong. It could be wrong because a) it’s not aerobic! or b) you could go harder and it would still be aerobic eg My partner runs too fast for aerobic workouts and I probably run too slow for aerobic workouts, we both should know better.
Being a personal devotee of ‘zones’ over the last decade, my advice to myself and others eventually boiled down to using a method of perceived exertion to determine your true Zone 2 against what the models came up with. Not very #scientific.
HRV Logger App
The HRV logger app does what its name suggests and I would class it as a ‘technical app’ suitable for those who wish to record and analyse HRV rather than an end-user tool to estimate LT1. That said it DOES estimate LT1 in realtime. It costs $10/£10
You can see in the image to the right the two settings that need to be made.
Real-Life Usage Protocol
In my tests so far I have found that the Alpha 1 value can suddenly change when running outdoors at less controlled speeds. Whether that is due to the sensitivity of the reading, my lack of pacing or the sampling of the data I’m not sure. The practical recommendation, therefore, is to test this in a highly controlled indoor environment. Have a proper warm-up and progressively increase efforts eg 10w every 2 minutes. You will hit the 0.75 mark relatively quickly as this is not a maximal test.
- Only the Polar H10 is recommended
- Only a BLE connection is recommended (ANT+ is more likely to lose packets and impact HRV quality…this is NOT an issue for normal HR when ANT+ is fine)
- A gentle ramp test is recommended, increasing efforts every 5 minutes (4-6 minutes)
My best result was 6bpm lower than I expected and that was after a 3.5-hour workout the previous day, so I was certainly fatigued.
Fascinating observation today. After a comment by @altini_marco on possible ANT+ vs bluetooth differences in signal quality I did a comparison of ANT+ vs bluetooth via the same H10 and Garmin watch. Big difference noted, updated my FAQ with more details of this. pic.twitter.com/NehHuc29TT
— Bruce Rogers (@bjrmd) February 25, 2021
Please also note that running outdoors will experience impact forces which may distort results (see this research). so a treadmill or bike trainer might be best to test this out for the first few times.
With the most accurate HRV-capable chest strap ie a Polar H10 or Polar H7 you can use Altini’s HRV iOS logger to mirror its display to your Apple Watch. The app/watch shows your Alpha1 change in real-time as you workout. Once your Alpha 1 falls to 0.75 then that is your LT1 point. It should correspond to a power or heart rate value that you can then use in your training zones
Once your Alpha 1 falls to 0.75 then that is your LT1 point
This will probably only work in easy ramp scenarios ie not for interval workouts.
Some of the researchers are looking into the possibilities of defining LTHR/LT2 by similar methods. That would be useful for those who train by HR but for those who train by power I think we have the algorithms we need to model FTP.
The HRV Logger app is not great as an end-user tool and someone needs to develop a Garmin CIQ app that uses the same algorithm (ping me when you do please) and similarly, some running and cycling recording apps should be googling away right NOW on how to do the math and how to incorporate this into their products. It’s public domain information, let’s hope someone has patented it as now it’s too late if you haven’t!
With a live metric, it may well prove possible to see how your threshold changes due to fatigue, caffeine or indeed throughout your workout or race.
I suspect the some sort of AI/ML tool will be needed to view your exertions in the context of your recent, full workouts.
If Alpha 1 is accurate then this could be one of the biggest ‘discoveries’ in recent years, one that can impact all of our daily training and its importance should not be underestimated. It could impact the training of EVERY person who reads this article.
This method has significant, potential benefits for endurance athletes who should be spending 80% of their training time performing aerobic exercise. It can easily be incorporated into your existing training regimes and could bring immediate efficiencies to your training.
This method has the potential to be incorporated into MANY consumer-grade apps and platforms to better guide those seeking weight loss, keeping them closer to LT1 and just below it.
Note: I paid for the app myself and get no commission if you buy it
- HRV Logger App
- Experiences at Runalyze (via @Rui)
- S Seiler & Tønnessen
- DFA a1 and exercise intensity FAQ on muscleoxygentraining.com by Bruce Rogers
- Gronwald, T., Rogers, B., Hoos, O.: Fractal Correlation Properties of Heart Rate Variability: A New Biomarker for Intensity Distribution in Endurance Exercise and Training Prescription?, Frontiers in Physiology, 11, p. 1152, 2020 doi:10.3389/fphys.2020.550572
- Rogers, B., Giles, D., Draper, N., Hoos, O., Gronwald, T.: A new detection method defining the aerobic threshold for endurance exercise and training prescription based on fractal correlation properties of heart rate variability, Frontiers in Physiology, 2020 doi:10.3389/fphys.2020.596567
This content is not sponsored. It’s mostly me behind the labour of love which is this site and I appreciate everyone who follows, subscribes or Buys Me A Coffee ❤️ Alternatively please buy the reviewed product from my partners. Thank you! FTC: Affiliate Disclosure: Links pay commission. As an Amazon Associate, I earn from qualifying purchases.
48 thoughts on “DFA Alpha 1 (DFA a1): new training threshold discovery method via HRV”
Need an android alternative!
I think there are going to be quite a few alternatives fairly soon!
Best alternative at the moment is polar h10 hrv recording and kubios analysis in my opinion.
But ideally we need something with real time feedback, so I can test several paces in treadmill for example, and understand where’s the breaking point. Already have Polar H10, can I use Polar Beat for this?
Using polar beat and H10 i wasn’t able to export the hr data from the training. I’m not sure if it available only available with polar v2+H10 training recording
Use https://runalyze.com/ it will do the same estimation based on https://colab.research.google.com/drive/1GUZVjZGhc2_JqV-J5m1mgbvbiTBV9WzZ?usp=sharing by Marco Altini.
Based on those papers:
Use steady-state or incremental easy effort every 4-6 minutes to get accurate estimation
thank you, those papers are already linked in the article
Where in the ui is this function? I just registered today so not sure I know my way around the ui well yet as I can’t find it
Practical application still need refinement: https://blog.runalyze.com/features/aerobic-threshold-estimation-based-on-hrv/
Managed to pick this up for £5 on the Apple store. I already had paid for HRV4Training and there’s an HRV bundle offer on the store.
Do you use their app HRV4Training to get informations and data about your training performance ? It seems that they are trying to deliver quite a lot with a good sensor (chest strap ) and an app. At the current state they seem to be in competition with GARMIN ( which delivers a lot of data ) and Polar ( which has less parameters but pretty good ). What’s your take about this “competition” based on yours experience with Polar, Garmin and maybe hrv4training?
i already have an oura ring and polar h10 (for hrv measurements ) and hrv4training app ( which I’ve used for a little amount of time) but I’m still trying to decide what platform to use for training analysis and therefore which sportwatch to buy. If the combo hrv4training + oura is a good combo then a coros pace 2 should be enough to log training and send it to strava (and therefore to hrv4training). Thanks!
yes that is a good combo
not sure hrv would get sent with that route
Yeah the hrv data during training will not be available. But for hrv4training i can use the data provided by oura ring. Still garmin platform or polar should be a better choice for their hrv exporting but also the price is quite high
garmin you have to set the hrv on , on the watch. you can directly export from the watch.
oura will only give you resting HRV that has any meaning.
This is fascinating. I’ll eventually try to figure out the algorithm for FIT file analysis afterwards.
I’ve already mostly figured out how to calculate respiration-rate from HRV data for older Garmin models that do not have that feature like the Fenix5/935 (which Garmin could very easily backport with almost no effort but refuses to do so)
Open-source info here if it helps others: https://forums.garmin.com/developer/fit-sdk/f/discussion/245469/-
By the way you can also calculate LT pace from known vdot, it’s a simple formula.
$ltpace = (1609.344 / (29.54 + 5.000663 * ($vdot*0.88) – 0.007546 * ($vdot*0.88)^2)
Also, a half-marathon PR on a flat/windless course is likely LT pace but that’s obviously not something that can be done too often to check, lol
if realworld LT varies by fatigue and caffeine (for example) then a generic calculation wont reflect that, no?
Right, using vdot would be static based on best effort at that time. It’s a limited estimate but it’s also a “sanity check” to see if HRV derived value reflects reality in general. vdot doesn’t use HR, it’s effort based so it’s limited but an alternative when less data available.
It’s important to realize that optical based sensors cannot do HRV unless the subject is perfectly still (like sleep), it can’t capture the timing between beats with enough resolution otherwise. Have you tested/seen any optical straps that actually attempt HRV while active?
BTW that muscleoxygentraining site you link, whomever that is, they are genius. Reverse engineering firstbeat stuff takes some smarts, been trying for awhile myself but they really got it. I need to read everything else on their site.
(oops I think that link was from the runalyze site, so you indirectly linked it by linking to them lol, still lots of unusual great tech insight there)
these? I’ll add them above
DFA a1 and exercise intensity FAQ on muscleoxygentraining.com by Bruce Rogers
Gronwald, T., Rogers, B., Hoos, O.: Fractal Correlation Properties of Heart Rate Variability: A New Biomarker for Intensity Distribution in Endurance Exercise and Training Prescription?, Frontiers in Physiology, 11, p. 1152, 2020 doi:10.3389/fphys.2020.550572
Rogers, B., Giles, D., Draper, N., Hoos, O., Gronwald, T.: A new detection method defining the aerobic threshold for endurance exercise and training prescription based on fractal correlation properties of heart rate variability, Frontiers in Physiology, 2020 doi:10.3389/fphys.2020.596567
Do you happen to have some Hexoskin voucher code lying around?
I don’t especially want to go into that here. Race/Pace predictors have perhaps moved on a tad and also when using HR or power for training I need convincing that FTP/LT2 always predicts lt1 for everyone. hence if we can simply, repeatedly measure it during low level workouts it will be awesome.
apologies if you’ve seen/linked this already but the HRV-Logger author also made a free “colab” available to compute DFA and alpha1 from FIT files! very exciting, I have to play with it later:
Hi – I only have a Polar H9 at my disposal and my results so far have been “mixed” to say the least (I reach the .75 threshold way way sooner than I would’ve ever expected having trained by HR for over a decade). Can you elaborate on why the Polar H10 is recommended? I thought the H9 was identical but without the record storage ability.
The H10 over BLE was the one recommended.
I would assume they were highly similar or identical internally…but I don’t know.
yeah not getting great results from my tests with the H9, I can’t believe it’s the HRM though. Here’s one example, assuming outliers are ignored, the best I can derive from this is my AeT is between 131 and 98 bpm.
i think you need to do it in controlled conditions.
slow ramp stationary bike/treadmill…that kind of thing.
Has to be steady state for at least 4 minutes at a set level of effort in order to get your heart rate to level off to get good values
yes, as above Marco said that a gentle ramp test is recommended, increasing efforts every (4-6 minutes).
The recommended straps seems to be more cause that is what they have had access to to check. Outside of not being optical cause those can’t do it it’s hard to say for sure what straps are good and which aren’t. Most people and reviews only look at heart rate data so straps sometimes tweak what they broadcast to give better looking heart rate which can make hrv data messed up
the sports lab based people tend to use Polar in my experiences.
Historically Polar straps were significantly better and since they worked well in the past there has been no incentive to see if they can trust other straps.
The interesting thing to notice in the “which strap is better for HRV” and the use ble over ant to get better data is the implication this has on garmin device. Most firstbeat functionality on garmin devices uses hrv so if you’re not using a hr source that gives good hrv data and if you use ant to connect (the default for garmin devices for those that could use ble) the first beat functionality won’t work as good as it could. So the advanced functionality on the garmin devices could work much better but no one does these tests in any of the reviews. Would be interesting if someone who had two devices pairs one to ble and one to ant from a dual strap and see how the metrics differ.And maybe one optical and one chest strap which should show an even bigger difference.
historically: yes, probably. plus many of the lab tools must have been ble-only.
there is also the consideration that Polar see the H10 as better than the H9 because of the strap. It physically holds its position better (nobbly bits) and, IIRc, has more/larger sensor pads. So the CHANCE of picking up all the data correctly is probably increased. (I covered that in my h10/9 reviews). this probably makes little difference to HR but maybe it matters for HRV? along the same lines that it matters for HRV that ANT+ might drop the odd packet here and there
Marco specifically raised the garmin/ble issue to me and i pointed out that most garmin devices will pair by ble. i don’t know where he will take that info…others on twitter talking about dfa1 seem to now be aware of that, so…
comparing the devices for changes to FB metric – hmm yes it would be of some interest. I’m not going to volunteer tho!
I guess the take-out concern is this: at least 95% of garmin users of FB metrics will use ANT+ and/or oHR. is their data enough for meaningful results?
For h9 straps, he doesn’t seem to think they have a downside https://twitter.com/NeilPugh5/status/1371458544596307973?s=19
that’s what I would have thought.
other than the properties of the physical strap.
the hrv tools site says
Android support is out now for the logger app
Unfortunately doesn’t work for me. I was doing an LT test on a treadmill using lactatedge device. My thresholds are 153/172 with HRmax 186, but I was hitting 0.75 value of DFA a1 at about 130…
LT1 cannot be 152 at LT2 == 172 and HRmax 186
2 and 4 mmol lactate does’t mean that this LT1/LT2 thresholds.
Just noticed there’s a Garmin IQ app for this: DFA Alpha1
Has anyone tried it? I’m sick right now, so I’m unable to try it at the moment.
i think this is the one and it should be ok. try ai endurance tooo…theres lots of interesting stuff going on there
Only compatible with Edge!? Can’t try it then…
Gold Standard for ambient ECG monitoring is multiple channels. The only one that is FDA approved that is comfortable is Peerbridge Health. A single channel ECG is NOT is not the standard for HRV with movement.
but for your average Jo at 8am when they wake up…
Comments are closed.