I have implemented both of these and will deploy Sat AM (GMT+2).
5 Likes
This is awesome 
Apparently the channel for running power is off. Will there be something similar for running like pace vs HR? Or running power vs HR 
?
2 Likes
I wonder if this linear relationship can be used to estimate FTP? The point where the linear relationship ends is your FTP. Almost correct for me 
Maybe garmin is doing something like that.
If the line is steeper after training, I find the training effective. If it is less steep, it suggests that you rest and recuperate.
The do something similar to get VO2 max and you can estimate FTP from that so quite likely.
Thanks for this David! Is there any way to change colors on the lines? Iâm red/green color blind so itâs pretty difficult for me to tell which season is which right now.
This is a great addition to intervals.icu!
However, as @Kosio_Varbenov already mentioned I also think that it would be more logical to have heart rate on the x-axis.
This is already given in its name: power vs heart rate. So power is the dependent variable which is usually shown on the y-axis.
1 Like
I was a bit worried about that. I will add an option to choose colour palettes for these charts.
I have swapped out the default palette for one that is supposedly colour blind friendly (all 3 sorts) based on this post: https://davidmathlogic.com/colorblind/
Wow, that was fast. Thank you David! This is so much easier for me to read.
1 Like
Cool stuff. Since it is HR Data the Indoor/ Outdoor part seems important. Cooling is so different and plays a factor to HR I guess. Now it would be cool to compare Ride vs Ride Vs Ride like this, you can set up a reference ride 2h Base ride for example do it every 6 weeks for example and compare them over time. Or you can do a Ride where you go in all Zones for a reasonable amount of time and do the same comparison over time. That would be so cool to have.
But for now, I have to load some more TSS to finish the week
1 Like
Hi David,
This is an amazing tool youâve created!
The popup is great, it shows the data of interest but the graph could be better.
I have one question and two suggestions that i hope youâll consider.
Question:
When this tool samples our data, what period of time must pass for a Power/HR correlation to be valid?
I am considering adding a very slow ramp up test every quarter to access stroke volume changes.
Do I need to ramp up slow enough that I spend 60s at each bpm or is 10-20s enough time?
Suggestion 1:
In my mind, it makes more sense to display output (wattage) on the y axis, given an input (HR) on the x axis. I think most users would want to increase power at a given HR rather than lowering HR at a given power. Please correct me if Iâm wrong, I image professionals may prefer efficiency (lower HR) as the performance metric because they have already reached their peak muscularity at race weight (w/kg).
ie. tracking improvements in stroke volume is measured with a static heart rate therefore wattage is the only interest, as shown in the popup. Vertical increases make more sense than a shift to the right (appears lower).
Suggestion 2:
Adding a cadence filter to the tool would be hugely beneficial for users that spend a season with/without torque intervals. High cadence is metabolically taxing vs muscular loading of low cadence work, so low cadence / high torque intervals will have a huge impact on comparisons.
eg. User cycling @ 60% FTP alternates cadence every 5 mins (50,75,90). Same fitness and stroke volume but different metabolic cost. HR may be 122bpm, 126bpm, 132bpm.
Great point. I think there is value in the current configuration but less so for aspiring / building cyclists.
Keep this one and add âHR vs powerâ for the next update.
Great work David!
Tx. It breaks the ride into 1 minute segments from the start of the ride. So your ramp test would need to have 60s steps and start on a minute boundary. But you probably donât have to do a specific test for this as there is likely enough data across all of your rides.
Quite a few people have asked for the axis to be switched around. I will look at adding that as an option + cadence filter.
2 Likes
Do you think this could be used to ascertain relatively accurately the LT1 and LT2 points by deflections in the linear power/Heart rate line
Hmm. Not sure actually. I googled a little and found a paper that says no:
3 Likes
Thanks Just a thought really as it would save an awful lot of testing if it did.
1 minute segments sounds kind of short- I wonder how this graph would look if it were, say 3-4 minute segments.
EDIT: the rationale being, if I did a ride at easy endurance and did a 1 minute max interval every, that ride would give inflated PWR for a given HR, or generally for efforts that are usually short
You can see the 1 minute segments on the ride power page on the first 2 decoupling charts. These are adjusted for HR lag which helps with your 1 min max effort case. Those 1 min max efforts will also have high power and fall into the non-linear part of the curve. So for the under threshold range is probably still good.
I agree, I donât mean to criticise, just brain storm- this stuff is exciting
I was trying to do something similar a while back. I had some ideas about how to get good data, but could not verify due to lack computing skills⌠One idea was to do linear regression on a per interval basis and only accept values, where the slope of HR and PWR data was in a limited band, as filter
Tx for the input on this topic! It discards minutes that are not 100% moving but thats the only filter currently. Open to any ideas on how to clean up the data more.