Vo2 Max accuracy

On intervals.icu there is a listing in the power menu that lists estimated Vo2 max. Mine currently according to this is 49.4 which when checking is exceptional for a 65 year old (highest age on the chart I looked at) and I am 69! ( I have seen as high as 50.4 before!) My last 4DP test showed my weakness was my MAP (Vo2 max) presumably in comparison to my FTP result. So, has anyone done a lab Vo2 max test and compared it to the estimate in intervals.icu because that would be interesting and pleasing if it was pretty accurate.

I would imagine the third party algorithm that intervals.icu uses is based on a regression analysis of lab tests.


I see a lot of variation across different systems, for what it is worth. I have three systems that provide estimates, though do not have a lab test to compare:
intervals.icu: 52.8 for past 42 days and 59.1 for 2022 (cycling data for inside & outside)
Garmin Connect: 59 for November outdoor rides, and 60 peak for 2022 (cycling outside only with power meter)
Apple Health 47.2 for Nov with 50.6 peak for the year. (walking, cycling (without power), yoga, etc.)

P.S. Age 52.

VO2 max definitely depends on the sport and how it is measured. Running and cycling for the same athlete would have different VO2 max values.

So the Apple Health used heart rate, and the others used power?
Looks like the 2022 cycling results were identical for intervals.icu and Garmin.

It would not surprise me if estimates using HR and power would vary, or the exercise mix on the Apple Watch gives different values.

intervals and Garmin both have access to heart rate & respiratory rate from a chest strap along with power data from cranks or Tacx 2T trainer. Garmin Connect algorithm also has access to other typical cycling data via speed sensor, cadence sensor, GPS locations. Speed is used for some VO2Max estimation methods. On the Garmin Connect report for the VO2Max it refers to "Advanced heartbeat analytics by Firstbeat® Technologies Oy. " Looking at their website, the following link suggests 95% correlation to lab results using only heart rate and respiration rates. They also have some algorithms that use heart rate and speed, for a given activity, using stable portions of time from the exercise.

Here’s one link to their algorithm using only heart and respiratory:

Apple Health would have access to both heart rate and respiratory rates, but for all kinds of activities beyond cycling. Thus, it isn’t surprising that it is the farthest away from Garmin and intervals.

1 Like

I had read the First Beat white paper a while ago, and yes they use HR. I do not know what intervals.icu uses. I asked because I understood you (incorrectly it seems) that you did not have a HRM for the Garmin results.

Looks like intervals.icu takes your 5 minute max power for a time period, assumes this was a maximum effort, then applies a model based off of 46 road cyclists:

The paper and video also describe a more accurate model when performing a graded exercise test.

Apple health just uses walking and running workouts—nothing else. For me, I don’t run and my walks are usually leisurely so the number is 12+ points lower than Garmin’s estimate.

1 Like

Several questions then.

If you are correct about intervals.icu, Garmin is using a different algorithm. I would imagine that intervals.icu did not want to pay First Beat. That then raises the question of the accuracy of any model. Also, as you noted, a 5 minute max power is not equivalent to a ramp or step test.

The algorithm in the paper and the video are different. They were done in different years, and have different resulting equations. You could probably take your results, calculate with both equations and see what results you get.

Of course, the issue is how representative the samples used in these studies are to us. I presume SYSTM uses their own algorithms based on their lab tests in creating Half Monty.

Modeled on 46 cyclists? That seems like a pitifully small sample size. I would guess the accuracy rates at the older end of their scale suffer the most.

1 Like

Interesting discussion.

I didn’t realize that Intervals.icu even estimated VO2max. As a 4DP test “Pursuiter” my 5 minute power is high in relation to all my other metrics, so it wildly estimates this at 64.9(!) - I so wish.

About every two weeks or so, I run the “Fitness Test” that Polar offers in their Polar Flow phone app in conjunction with their HRM. This “test” just requires that you have been rested for a bit and comfortably lying down. They also recommend that you run this at the same time of the day for comparisons. I tend to do mine on Sunday evenings after dinner. With this app, I get values from 54-60, but typically 57… I’m also 52 years old.

Also, Intervals.icu over estimates my eFTP, also based on my 5 minute efforts. Was good to know based on 4DP results why this was the case.

1 Like