I’ve been trialling Athletica AI and I really like the app and the concept behind it.
But there is an issue, at least for my profile, that makes the user experience a bit bad. Which is the completely off critical HR calculation.
I understand the garbage in garbage out concept, but in my case I have been using a chest strap for the last 3-4 months and I did the test workouts last week and the system keeps computing a critical HR of 183, when my max HR is actually 182.
If I manually override it then everytime there is an update the system reverts back to the 183. This makes the HR zones being completely wrong and with the setting of letting the AI decide the best way to submit the workouts to Garmin (the last ones where submitted with HR targets) makes the user experience really bad.
Is there a way that someone could have a look and understand the reason behind this? e.g. Garmin with the same historical data computers max HR of 182 and threshold HR of 171.
Sorry about your bad user experience. As I explained to you via our support, the system is doing exactly what is designed to do: automatically scan over your data to determine your critical heart rate and pace/power. It looks at the data the past 6 months finds a workout where your heart rate has stayed elevated over a certain period and determines critical heart rate based on that data. It only sees it as the maximum recorded data - whether it is coming from optical, or gold standard chest strap. In your case, as you already explained to me, that you started you use a chest strap the last 3-4 months. What is most likely happening is that there are optical hr data “messing up your data pool”, skewing your true, real critical HR.
I understand it is not a great user experience, and we can certainly think of some way to control check data quality. But the system is doing what is designed to do. Garbage in - Garbage out.
If you manually override the Critical HR - it should stick, and if it is not sticking - let me know and we can have a look.
Once your erroneous data falls outside the 6 months data range, and you have better quality data inside the 6 months, your critical heart rate will be more accurate. OR as I explained via customer service, you can manually delete those workouts that recorded higher than normal heart rate to clean your data (check your Garmin Connect account for those workouts so you know the dates).
Apologies for bad customer experience, we are not perfect but we do strive for excellent customer service.
First many thanks for the quick responses and great support you are providing.
I checked in Garmin and I started using the chest strap 17/May. I have manually deleted all from 1/March till 17/May. Is there a way to force a recalculation?
As to the override the critical HR, the one you did to 171 got automatically overriden after my last workout was uploaded.
Finally for the AI workout upload my last HIIT MIT workout was all uploaded with HR targets and not just HR targets for Z2 and pace targets for the others
Great! Thanks for clearing out the old data. the system checks everyday, so you may see a change tomorrow. Keep me posted if you still see CRHR not sticking to 171.
So check your in your connection settings how you have asked the system to send to your Garmin. Hit “Resync all future workouts now” to force.
Hope this helps,
MJ
I confirm that after resync the target type havs been updated for the non-aerobic workouts.
One thing to note is that if in the user settings we set that “my device supports power” even we if set that the preferred running prescription to pace the AI uploads the workouts to Garmin with power targets.
Once I deselected the option “my device supports power” after resync the workouts were created with pace targets.
I will report back tomorrow if the critical HR is updated from 183 after the overnight update.
Today AAI suggested updated threshold values (for cycling). The suggested CP was completely reasonable, but the THR was suggested at 177. While it has been at ~166…169 for the last years. In fact, I would argue that my 20 min test last week confirmed that quite well (i.e., 0.98 * (20 min avg. HR) for that test was 169; I understand this is not how AAI determines it, but it is still a decent estimate;-)
The suggestion of 177 came after VO2max-30/30-intervals, in which however HR never went to 177 nor above (max HR = 174 IIRC).
Would be nice to get the HR calculation more robust;-)
There’s actually nothing wrong with the algorithm — it’s working exactly as designed. What happened here was simply messy input data (from older or inconsistent sources) confusing the system. Once the data stream is clean, the calculations line up as expected.
Thanks for flagging it so clearly — it helps us spot where data quality can impact outcomes.
@Marjaana my critical HR updated again automatically to 183. This is so annoying.
Can someone please really check why this is happening with my profile and fix this. I don’t have any workout in the app since Jan this year where my max HR gets to that number, so I can’t understand how the critical HR computed by Athletica is 183.
This can’t be about bad data. There is something definitely wrong with the calculations and model used by Athletica.
Thanks @RuiMendes
I understand it can be confusing. Please rest assured that the model is solid. I’ve flagged your HR constantly changing even with your removal of bad data with our back end team. The manual change should stick, and when it doesn’t it is usually an issue that can be fixed.
Hey, @joku sorry late reply here. I’m not here to argue I believe you when you say your threshold has been around 166-169 for the last years. Usually experienced athletes like yourself intuitively know where the threshold sits. When it doesn’t - my first question is: is the data good quality? Sometimes there’s a bug that the team needs to take care of.
The question remains @joku your CHR is set 168bpm. If that is a manual override and it reverts back - please let me know via support so our team can have a look shortly after it changes. Thanks!
MJ
It’s fine. I’ve already logged the issue with the team and I know where to find you IF HR changes again, then please do open a support ticket. But for now - I’ve got you.
Yes:-) Really, there is reliable chest-strap data pretty much for every training and there is (very comparable) Watch data for most commutes (some also have chest-strap data).
I am aware of this and I have been analyzing P, but not HR, using critical power models (as implemented in i.icu) for years myself. I am aware of the differences and similarities.
Yes, I manually set it when starting with AAI. And I set the “update CP”/etc. to “ask”.
It does ask, but only for CP updates and CHR updates at the same time, so I cannot accept the new CP without also accepting the new CHR… And, more important here, the suggested CHR are really out of place (some were >20 bpm away from the 168, IIRC always way too small, Z2ish, very close to VT1-HR).
PS: I do vaguely remember that it also updated it nevertheless I did not accept it, or without asking, but I am not sure anymore – thus there is nothing to do here in that respect for now. Will report in case this really happens (again).
@Marjaana Shall I manually update my critical HR manually to 171 or is best to leave it as is for the time being for the backend team to have a look at it?
I believe you when you say your threshold has been around 166-169 for the last years. Usually experienced athletes like yourself intuitively know where the threshold sits. When it doesn’t - my first question is: is the data good quality? Sometimes there’s a bug that the team needs to take care of.
IF HR changes again, then please do open a support ticket. But for now - I’ve got you.