šŸŽ‰ šŸŽ‰ šŸŽ‰ Introducing Adaptive Training! šŸŽ‰ šŸŽ‰ šŸŽ‰

Will it use heart rate as a signal when determining that a workout is too hard?

Because we have multiple inputs that are proxies for fatigue & freshness: power stability, heart rate (potentially including HRV), and even cadence changes. And Firstbeat seems to know how to grab breathing rate from heart rate, so that could factor in at some point as well.

I’ve seen people muse on the Reddit Velo subreddit on whether failing an interval / going to failure is the only way for it to recognize that you’re pushing too hard; realistically, running ragged on heart rate and your cadence going all sloppy could be signals it could detect too.

Super excited to see the first iteration of this, thrilled it’s here, and am going to love seeing where it goes!

I think this is really promising, though a little odd that the launch is not correlated with immediate (albeit limited number of user) access.

I don’t think so, from what I gather. It will be purely the subjective survey response, and the actual workout target vs reality comparison.

Nate did mention HRV, but that sounded like further out goal for implementation.

time frame 3 months, 6 months, 2022?

No, but the RPE measure is more reliable anyway

no surprise…has to be an offering, but I like what Nate says about bringing other stuff. Being able to do block periodisation etc would be great, and it would be excellent to see results from all the data comparisons

Most ML specialists would suggest that several input variables (HR, RPE, breathing, HRV, power, cadence) are superior to one. In this case, I would imagine that RPE can be weighted quite heavily, but if they’re leaning on ML, I wouldn’t expect that to just stay with that factor.

What sort of email did you receive when signing up? I just received an email merely acknowledging that I will have priority access. Was yours different?

I could have used this a few weeks ago when I nailed SSB HV at 300w and started stinking up the joint for sustained power build lol

Officially official: (per Nate in final questions in Ep 298)

  1. Thing 1 = Adaptive Training

  2. Thing 2 = Group Workouts

I’m so excited. Looking forward to being selected by the AI overlords. :sweat_smile:

@ambermalika Congrats!

what happens if I am well into a plan already and then switch to the adaptive one, what happens with the plan I am on?

Does it just change to adaptive and learn from my previous performance on my current plan?

That’s what I’m thinking, also the plan has been working fine for me.
But I’m pretty stoked to have a even better personalized plan based on my goals.

I think, IMO, we will still set dates for races/goal events and then have same outline of periodization. Then while doing each workout they will adapt to how you’ve been performing… idk :man_shrugging:

Yes, key word is soon. I believe we’re all getting the list and will be granted access in waves.

This is terrific news and I can’t wait to try it out. Hope the Adaptive training will be able to take into account Master’s riders (like me) who may need a bit more rest than someone younger.

On another note, Dylan Johnson’s critique of the TR plans might have been quite fortuitous in terms of timing with this announcement…I wonder how many new views he drove to TR’s website, where someone initially thinks ā€œTR’s plans suckā€ only today to be presented with ā€œTR’s plans are adaptive!ā€

Am I right in assuming that this will handle the ā€œmasters’ plansā€? I.e. suggest recovery weeks as appropriate, suggest a slower progression, recognise lower Anaerobic & VO2max levels and similar age related issues?

RPE isn’t going to show your HR decoupling towards the end of a workout.

It will respond to every athlete individually. So rather than put you in a ā€œmastersā€ bucket, it looks at YOU and your performance, and trains you in a way unique to your needs.

I am interested in this. However, as a multisport athlete, I can only assume this will not be something that I will be able to utilize fully, as I am not confident that a ā€œfull analysisā€ is possible right now.

I would love to test this out though