6

Sports watches can reportedly give an estimate of one's VO2max. Tracking changes in this number is useful for training (even if the absolute numbers turn out to be unreliable).

Can I track (improvements in) VO2max using just the data from a heart rate monitor, a cadence sensor, a speed sensor, and GPS data (distance, altitude), plus additional software on mobile or desktop? What does a sports watch have access to that could not be done with these data?

5
  • 3
    I am not familiar with all sports watches so this isn't an answer to your question but I'm slightly familiar with Garmin watches. For cycling activities, Garmin watches need a source of power data in addition to HR data to get an estimate of VO2Max; however, even with a HR belt and power data, the Garmin estimated VO2Max is suspect. For running, Garmin watches don't need a source of power data so in theory they could do without it for cycling--but the error would likely be even worse. That bodes poorly for other brands of fitness watches.
    – R. Chung
    Sep 8, 2023 at 14:08
  • 1
    Also not an answer to your question but as an aside, VO2Max doesn't "move" as much or as quickly to training stimulus as some other metrics. In general, it's nice to have a metric that responds to training stimulus quickly and can show smallish improvements. Watches often show integer values for VO2Max so a 1 pt. change is about 2%; a 2% change in FTP is much more noticeable. Runners may not know their VO2Max but they usually know their mile pace: a wristwatch measures smaller changes in running speed.
    – R. Chung
    Sep 8, 2023 at 16:06
  • For interest's sake, there is a portable VO2 analyzer, which directly measures O2 consumption but doesn't require a lab. I don't know of any reviews of its accuracy. It's also a bit expensive. vo2master.com
    – Weiwen Ng
    Sep 8, 2023 at 18:37
  • @R.Chung Re: "VO2Max doesn't "move" as much or as quickly to training stimulus as some other metrics." What can I track to show improvements, instead of VO2max?
    – Sam7919
    Sep 9, 2023 at 10:44
  • 3
    @Sam Elapsed time over a known course is a pretty classic and easy thing to track that doesn't require either a power meter or even a HR monitor. Alternatively, if you had a power meter, there are several power-based metrics you could track; for example, I happen to use CP and W'--but there are others.
    – R. Chung
    Sep 9, 2023 at 15:51

5 Answers 5

4

Can I track (improvements in) VO2max using just the data from a heart rate monitor, a cadence sensor, a speed sensor, and GPS data (distance, altitude), plus additional software on mobile or desktop?

No. You need a consistent source of power meter data (along with an HRM).

With consistent power meter data, something like Garmin connect will be able to estimate your VO2. In the case of Garmin Connect, it will alert you when the estimate changes.

2
  • 1
    That's helpful to know, but at the risk of making the question a general sports question rather than a cycling one, it's worth mentioning that VO2max may be a parameter of the athlete, no matter the sport. And so potentially we could leave the bike (and power meter) behind and still seek an estimate, perhaps with just a pair of sneakers (trainers) or a swimsuit. Or is it the case that the estimate is very much sport-specific?
    – Sam7919
    Sep 8, 2023 at 18:53
  • 2
    @Sam VO2Max is sport-specific: the same athlete, when tested in different sports, will produce different VO2Maxes. Running and XC skiing tend to elicit higher VO2Max than cycling.
    – R. Chung
    Sep 8, 2023 at 20:48
3

Unlike running, bicycles are inherently very efficient, therefore with cycling the power lost to speed related items (height gain and drag) varies much more than running. Faster speeds mean wind (power required to maintain a speed is roughly a square of air speed) and hills (gentle inclines) affects results far more than running. Garmin can guestimate VO2Max running provided the runner is on a moderately flat course and its lowish wind. On a cycle, a slight incline or slight wind can easily make a big difference to required power to maintain a speed. Therefore, cycling Vo2Max can only be guesstimated if actual power is measured. Running varies much less and therefore it is possible to guestimate Vo2Max with useable consistency.

A good example that is easy to understand is imagine if you were guestimating Vo2Max with two clones riding a bike. One was in the lead, the other drafting the entire ride. Same speed, same distance, same height change yet very different Vo2Max. If you did this with runners, you would get a different Vo2Max, but it would be much closer.

3
  • 3
    power required to maintain a speed is roughly a square of air speed It's worse than that. Drag and therefore force is proportional to airspeed squared. And power is defined as force times velocity - so power is proportional to speed cubed, assuming no wind. The calculations get more complex in any wind, but there will still be a portion of power proportional to velocity cubed. Sep 8, 2023 at 22:33
  • IIUC, a cyclist climbing some known grade, in still wind and without benefitting from drafting, will have the extra workout reflected in an increase in heart rate. It's just that no one has figured out yet how to come up with a VO2 max estimate from HR data (and altitude data). As you mention, determining whether the cyclist is drafting or is facing head wind is not possible with current tools/sensors, and so power meter data remain necessary for either.
    – Sam7919
    Sep 9, 2023 at 2:25
  • @Andrew - thanks, I was not sure about it being squared or cubed- as you say, it is complicated to work out and I did not want to overstate the problem of wind on the power vs speed of a cyclist.
    – mattnz
    Sep 9, 2023 at 4:23
3

if you only wants to see if there's any improvements in your VO2 max, then it is possible as the absolute number is not your main point, it is your relative (comparison) to the previous VO2 max on the same device. as to how, I also didn't have power meter back then, the similar setup as you (cyclocomp with GPS, cadence, HR). I'd go to a climb (preferably under 10 minutes, pretty steep so you won't deal with much aero drag and can go all out). try to go all out for 4 minutes (ofc. after warming up first get the HR to zone 3-4; [use HR zone calculator if you don't have your HR zones 1). then hit the climb all out for 4 minutes, pace your effort so your HR is at zone 5 at the later half of the climb and sustain it until the 4 minutes is up. the do the same effort after some training, and see the improvements of speed and distance from your 1st and 2nd effort. or, you can do the same climb with the same avg speed, then compare the HR differences between 1st and 2nd effort (it should decrease).

1
  • 1
    Good point. Indeed, if I'm drafting while doing any measurements, then I'm basically "cheating" (myself). The relative improvement is all that matters. Now the questions are 1- why does climbing for 4 minutes in HRZ 5 adequately capture my VO2max, and 2- can I do the same using something else that I do anyway: ride to a nearby town (about 100 km roundtrip) and track my average speed, week-after-week, month-after-month (and year-after-year) ?
    – Sam7919
    Oct 12, 2023 at 15:13
3

First, I'll provide one answer to the question. After the line break, I will raise the question of what your goal is - you may not need to be interested in your VO2max specifically.

If you at least have power data, it is possible to estimate VO2max from a 4 to 6 minute maximum fresh effort. If you are on Zwift, they already do this with your existing power data. Based on a 5-minute max effort where I averaged 295W, Zwift estimates that my VO2max is 57.3 mL O2/kg/min. Basically, an effort around this duration should put you at or near your VO2max.

If you Google, you may find some formulas in peer-reviewed literature or elsewhere. Andrew Coggan, a respected sports scientist, posted here that you can use this equation using the mean power from a maximal 5-min (I think) effort:

VO2max in L/min = 0.0108 x power (W) + 0.007 x body mass (kg)

Note that the formula provides liters per minute, not mL, and it's not normalized to body weight. When I input my 5-min power, multiply my result by 1,000 to convert L to mL, and then divide by my current body weight, my 5-min effort corresponds to an estimate of 56.47 mL O2/kg/min, close to the Zwift estimate. This study validated one estimation equation that used the maximum power from a ramp test. It also compared with some equations that use only height, weight, age, and gender. This one takes your 5-min max power in watts per kg, and I think that the training platform Intervals.icu uses it. If an estimation equation doesn't use actual power output, then you should interpret it as giving the average VO2max given certain characteristics. You should be careful taking that as an estimate of your own VO2max. This is like the max HR = 220 - age issue that I've written about elsewhere.

There are some assumptions that go into these estimates.

First, you have to assume your power meter is accurate. If you have a modern smart trainer or a good dual-sided power meter, you are probably good enough, but it is possible for a unit to come out of calibration. Of the major power meter brands, Shimano power meters are the only ones that have been tested to be systematically inaccurate, which is incredibly disappointing but is outside the question scope.

Second, I believe the formulae assume a fixed gross mechanical efficiency. That is, you're measuring watts generated at the crank or the trainer flywheel. Your body has to burn fuel to generate that power. It is not 100% efficient. In fact, one study I cited in this answer found that trained male cyclists average 21.9% gross efficiency with a standard deviation of 1.7 percentage points - that is, 95% of the population this sample represents should have GEs from 18.4% to 25.4%. Thus, Zwift is really reporting the midpoint of a range of possible VO2max values. Zwift's estimate has the caveat "Assumes a fixed cycling efficiency", and I believe this is what they mean. I suspect the GE assumption is baked into the constants in the formula, so I don't know how to re-estimate a plausible range for myself. Consumers can't easily measure GE. You could measure it in a VO2max test if you also measure power to the bike, but then you already have VO2max directly measured.

Third, I don't have a background in exercise physiology, but I believe this bit is accurate. We know that anaerobic metabolism contributes to 4-6 min efforts, especially at the start of the effort. If you are an anaerobic-dominant athlete, the formula-based VO2max estimate may be high. You can often get a general sense of your anaerobic capabilities based on how you do in sprints or very short max efforts (<2 min) in group rides or races. I know that I have good anaerobic power. So, my Zwift VO2max estimate is also biased upward, and I don't know by how much.


Also, readers should consider why they need to measure/approximate their VO2max. Are you looking to maximize longevity? Per the discussion in this answer, for most adults who want to maximize longevity and who don't need to get high athletic performance, it is likely sufficient to do long slow distance workouts. You can add whatever intervals you want, but it might not be necessary to improve longevity.

If you are a cyclist and you want to improve your performance in short-duration efforts, then you can track your 5-min power as discussed in that answer. It can be seen an imperfect but reasonable proxy for VO2max, as discussed above. Or you can just take it at face value.


A bit more about estimation equations and accuracy

This is posted mainly for my interest. To get an estimation equation, researchers gather a bunch of volunteers, then put them through ramp tests in their exercise lab, which gets them actual VO2max. Then they might fit a linear regression of measured VO2max on various characteristics, e.g. age, height, mean power from a maximal 5-min effort done on some other day, etc. That will spit out an equation like the one Coggan provided, or the 220 - age formula for max HR, or the formula below. Non-linear forms are also possible, as discussed in this post on TrainerRoad's forum.

That estimation equation has some prediction error. That is, you might be a bit above the predicted VO2max or a bit below. This blog post by Jem Arnold distinguishes between confidence intervals and prediction intervals. Some of the reasons that an estimation equation might miss your actual VO2max are discussed above.

You might also wish to consider the nature of the sample that the study used. The Sitko et al study linked earlier had an average VO2max of 61.13, standard deviation 9.05. That means most of the sample had VO2maxes ranging from about 43 to about 78. That is very fit compared to the population average. For your average person and possibly the most elite of elite cyclists, that study's formula is doing an out of sample prediction. There just isn't a way to be 100% sure that the estimation equation still holds up. There's not much an end user can do except try to find a prediction equation that's done on an appropriate sample. The FRIEND equation study (referenced in the Jurov article) used a more average sample - the average VO2max for men was 43.43, and for women it was 23.25. The FRIEND estimation equations for men and women respectively are:

Men: predicted VO2max in mL O2/kg min = 1.76 * ramp test power * (6.12 / body weight) + 3.5

Women: pred. VO2max = 1.65 * ramp test power * 6.12 / body weight + 3.5

Body weight is in kilograms, and the ramp test power is the maximum power achieved on the ramp test performed on cycle ergometer. A ramp test is a graded exercise test, where you pedal and the ergometer adds resistance at set intervals until you can't pedal anymore.

What's the prediction interval of the FRIEND equation? The Jurov et al study gave the graph below:

enter image description here

This is a Bland-Altman plot. This is one of the accepted methods to see if a prediction equation predicts VO2max values well. This was done by taking the measured VO2max and the predicted VO2max. Interpret the x-axis like the person's VO2max. The y-axis is actual minus predicted VO2max, i.e. how far 'off' is the predicted VO2max for each person.

For men, the prediction equation underestimates VO2max by an average of 3.6 mL O2/kg min (the line labeled CE), which the authors characterized as minimal. For most men, the equation's predictions should be within +8.32 and -15.52 mL O2/kg min of the measured value, where + means the FRIEND equation over-predicts, and - means the reverse. Those values are called the 95% limits of agreement.

You might look at the points on the plot and see a random smear around the center. This isn't some sort of Rorschach test, it's exactly what you want to see. You don't want to see some sort of trend. As an imaginary alternative example, say that the equation tends to over-predict among people with really high VO2maxes. In this case you'd have to consider that if you use it on yourself.

The FRIEND study didn't directly state the test format, in particular the starting power, how much power was added each ramp, and how long each ramp took. You may be able to use the maximum power value from a ramp test done on a smart trainer (usually that's multiplied by 0.75 to give an FTP estimate), but it might not reproduce the test protocol. If so, your result may not be accurate. I don't have a recent ramp test result to test how the equation predicts for me.

4
  • 2
    1. There are ways for regular riders to test whether a power meter is accurate. Maybe that should be a bicycles.stackexchange question. 2. The ACSM has estimation equations to convert power to VO2Max: my recollection is that the R^2 for these estimates for males is around 0.5, so reasonable but not great.
    – R. Chung
    Dec 25, 2023 at 21:47
  • 1
    BTW, here's a plot of data taken from Coyle's classic 1991 paper. Note that VO2Max isn't as closely related to "performance" as VO2@LT. anonymous.coward.free.fr/rbr/coyle.png
    – R. Chung
    Dec 28, 2023 at 4:01
  • Your FRIEND equation for females has an additive constant; does the male equation also have one? From your text, it appears that it ought to have one of about 3.6. Bland-Altman (more properly, Tukey mean-difference) plots sometimes tell you about systematic trends with level. Power at the end of a ramp test depends on the ramp steps and duration at each step. These all suggest that VO2Max estimation is fraught, which is why we often base fitness improvements on other, simpler to measure, markers.
    – R. Chung
    Apr 9 at 14:28
  • @R.Chung Correct, both gender-specific and the non-gender specific equations have an additive constant of 3.5.
    – Weiwen Ng
    Apr 9 at 20:26
2

It's worth pointing out that an otherwise popular product will not help. The Apple Watch will estimate your VO2max for only outdoor walking, outdoor running, and hiking, but not for cycling—not even if you have a power meter.

At least some Garmin Edge devices have for some time estimated VO2max — using software by Firstbeat.

Quoting a comment by Weiwen Ng, lest you miss it:

There is a portable VO2 analyzer, which directly measures O2 consumption but doesn't require a lab. I don't know of any reviews of its accuracy. It's also a bit expensive.

Other References

1
  • 1
    A complement: the Apple Watch also does "background measurements" of the VO2 max, not linked to specific workouts. Power meters are only supported from WatchOS 10 that will be released later this month (September 2023).
    – Rеnаud
    Sep 9, 2023 at 5:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.