terrymorse 

There are a few VO2max estimators out there that claim to be able to detect VO2max from a combination of power and heart rate, and they don't require periodic test rides at VO2max. They just watch every ride you do and work their statistical "magic".

I like the idea that you can just ride, because VO2max efforts are not pleasant. But are these estimator algorithms accurate? Are they precise?

Even if the absolute number is not accurate, watching changes over time might be useful. I started tracking VO2max generated by HRV4Training, increased my riding time in October, and the trend line pointed upwards somewhat:


And while the absolute numbers are nothing to brag about, it's encouraging to see an increase in response to more training. And hooray for not having to do VO2max fitness tests!

For what it's worth, the author of the estimation algorithm explains it in detail here.

Iride01 

I'm happy when mine goes up too. But I'm still not certain why I should be concerned with what it is. Since it's a moving target, what does it mean for me when it goes one way or the other.

Seems just another way for me to say that I need to ride more if it goes down. But if it is going up, would or should I ride less? Of course not. Regardless of where my VO2 max is, I'm still going to ride and do the appropriate training I might need to achieve whatever my goals are at that particular time.

Perhaps for a group of cyclists it might help weed out the persons that might need to be in another group. But that'll still needs to be cross checked with other data. Such as who has the longer time to the line.

RChung

Does HRV4Training use power and HR, or power and HRV for these estimates? I get 404's when I try to click on the links at the bottom of that page.

I get way too many HRV artifacts when I try to use apps that check my HRV during cycling. These apps seem to be okay when I'm sitting quietly on a chair or the edge of my bed, but not while exercising vigorously, so there's that.

I don't think that there's a single be-all end-all measurement of aerobic performance, and they all seem to move in the same direction (since a rising tide lifts all boats) so I tend to use the measures and estimates of them that are easiest to do and seem the most robust even if they may not be the ideal measure in specific instances.

hubcyclist

my VO2max estimate, whatever it may be, has zero impact on me and the way I'm training. I'm still gonna do vo2 blocks and try to go as hard as I can sustain on each interval, if the number goes up over time, i'm happy.

PeteHski

If your VO2 max really is 58 then you are at elite level by age.

terrymorse 

Here is a blog entry from the HRV4Training author, with lots of detail on the estimator algorithm.

https://marcoaltini.substack.com/p/e...-to-know-about

RChung

Thanks for the link. That was informative. Marco has a different purpose in mind: he'd like to get the world to do a little more exercise and get a little fitter, and that's admirable. Accordingly, he's sort of wedded to widely available technology like cell phones and HRMs.

We spent a lot of time looking for correlates of watts/HR back in the day, but in the end I figured since I had a power meter, I wasn't that concerned about watts/HR; rather, I was more concerned about watts/kg or watts/m^2 of CdA.

PeteHski

Interesting read. One thing I don’t understand with using HR to estimate VO2 max is the wide variation in HR max across individuals, regardless of their fitness level. For example a good friend of mine has a much lower HR max than me, but is also considerably less fit. How would this kind of model account for these big differences in HR across individuals? Would his lower HR simply give a higher estimated VO2 max? How would it know what his HR max and HR reserve was if there were no maximal efforts in the dataset? Is it just using a standard age-related formula for HR max? In which case it could be miles out for any individual.

However, I can see how the trend could be useful over time regardless of the accuracy of the absolute value.

FWIW Garmin is currently estimating my VO2 max at 55, which seems optimistic. Fitbit estimate is 42-46, which I think is only based on resting HR and personal stats.

RChung

Typically, VO2Max isn't attained at HRMax; typically, you're way past VO2Max at HRMax. Also, HR doesn't tell you about stroke volume or cardiac output, or mitochondrial density, or a host of other important physiological markers.

That said, Marco isn't looking for a "structural, causal" pathway from HR to VO2Max; he's looking for an equation that fits. That's a lot simpler, and amenable to the kinds of correlative analyses that ML is especially good at.

terrymorse 

I suppose it might be pretty good for an old guy. But that's not very satisfying. As Ned Overend said:

"If you think, 'Well, I’m doing pretty good for an old guy,' then you’re not trying to stay at the front."

PeteHski

I was only referring to HR max in the sense that the whole HR range is widely different between individuals of the same fitness level. So I’m not sure how the model can adjust for such individual variation. Especially if it doesn’t even capture an individual’s complete HR range. For example if it measures my HR at 150 for a particular power output, how does it estimate where 150 is relative to my HR max or my LTHR?

RChung

There's no age at which I've ever thought that. Usually I'm trying not to end up DFL.

RChung

I'm pretty sure from the description that Marco's algorithms don't really care about the complete range of HR; I'm guessing they look at something akin to watts/HR and how that varies across power and HR; the main underlying modeling assumption is probably that watts/HR isn't constant but changes smoothly. The algorithmic tricks are in finding the "right" lag or kernel over which to detect these changes, and modern ML methods are pretty good at brute-forcing this.

PeteHski

The Watts/HR I can understand. I just don’t get how you could convert that into a reliable estimate of VO2 max without understanding the complete HR and power range. But I guess that’s the challenge and limitation in not using a specific test protocol to capture the full range of HR and power.

I would imagine the VO2 max estimate improves if your data happens to include a full range of efforts across your power range. I don’t know what Garmin uses, but it seems to be accounting for my relatively high HR range. Fitbit only appears to use my resting HR, which is relatively high and I presume that’s why it estimates my VO2 max considerably lower. For example I had COVID about a month ago and my resting HR went up about 15 bpm and Fitbit reduced my VO2 max estimate accordingly.

RChung

I'm sure that's true: if a particular individual never gets anywhere up near, say, half of VT1 (let alone VT2 or VO2max) any kind of estimation based on that person's data is going to to have largish error. I'm sure that as the domain of watts/HR enlarges, the estimate of VO2Max improves.

PeteHski

That makes sense and I would imagine most users of this App would be feeding in a wide range of efforts as a matter of course. If I fed in a few weeks of my rides and workouts they would cover my full working range of HR and Power. I might check it out for interest.

The App I use for training (PILLAR) doesn’t provide an estimate for VO2 max. It focuses more on W/kg at the various key points.

RChung

I don't pay *that* much attention to V02Max because at the elite level (which is really the only time I see good measurements of it ) the relationship between V02Max and performance isn't terribly strong. I believe there's a selection effect where most elites have V02Max's that are already pretty high so it's no longer useful for differentiating among riders. It's sort of like SAT scores in the US: overall, they tended to do an okay but not stellar job in differentiating who goes to elite schools, but if you look at students at Harvard or Stanford or Berkeley, their scores don't at all predict their grades--SAT scores were used (in part) to select the students so whatever predictive capacity they had is already used up. In cycling, especially in elite racing, V02Max has already been used to winnow the riders so it no longer predicts performance.

Also, my V02Max sucks, so there's that.

Hermes

It is interesting to watch VO2 go up and down on my Garmin. And it says I have a VO2 of a 20 year old. I put the Garmin VO2 calculator into the amusement category..

Here is my take on devices and apps offered to us as training tools. I think accuracy matters. If I think the knowing VO2max will aid my training and increase my speed then I will go to a lab that has a track record and proven accuracy, put on the mask and do a proper VO2max test and pay up. I would then wait for a period of time and go back and take the test again and reset training if required. IMO, I suspect my accurately measured VO2 max with a proper test will be difficult to increase.

I have an SRM power meter on my road bike and I find it very difficult to improve my average power on climbs that I do a lot. It is very difficult to improve my time trial times on a course that I train on and race on. Average power for the TT always is about the same.

I find that improving a power metric by 5 watts difficult and a 10 watt gain over a season is massive for me. Maybe I am just too old or weak. My point is small changes in power matter and require accurate measurement to detect them. I suspect changes in VO2 will be similar. And a VO2 test in a lab is going to be brutal. One has to ride like a rabid dog and push beyond what one thinks is possible.

Terry may be able to relate to this anecdote. When I lived in NorCal, each Thanksgiving morning, I would race up Mount Hamilton with the Low Key Hill Climb series. We were lined up by Old La Honda climb times. Generally, that put me toward the back of the peloton. One year, I decided to ride with the 20 minute climb time group.

We took off and I stayed with them for the first 5 miles of the climb with 13 miles and two climbs to go. I looked at my power and I was riding way over my capability. But was I? My brain took over and decided if I kept up this pace / power I may not make it up the final 7 mile section. I let them go and it was like someone let the wind out of my sails. I finished with my usual time to the top. But I set the world on fire for the lower section.

I think we are all a lot more capable than we possibly imagine.

terrymorse 

I can certainly relate to that Mt Hamilton climb, as I try to do it once a week when conditions are okay. But I don't race it, preferring a steady tempo pace. That first 5 miles tells me how the rest of the ride is going to go. (BTW, the Low Key Hill Climbs are now "start whenever and post to Strava", no mass starts).

And your point about how our mind can control how fast you go -- totally! I remember entering my first mountain time trial, several years ago. I didn't have a head unit that would tell me my climbing rate, so I just went on feel, with a desire to not get passed by the next rider. Later, I did some calculations and found my 10-minute climbing rate was 4700 ft/hr, about 1000 ft/hr more than I figured I could do. I've never climbed that fast again.

That mental effect is one reason why I like tracking power vs. heart rate (and it's estimated VO2max) changes. It doesn't seem to be dependent on how motivated you might or might not be on a given day.

RChung

Hmmm. I disagree. I set my peak 5-sec, 10-sec, 15-sec, and 20-sec power on a ride right after a guy in a car buzzed me and my buddy. I knew he was likely to get caught at the next stop light and lit out after him. The real problem is that I didn't know what to do with him when I caught him.

[Edited to add] Though to be fair, I don't know for sure since I wasn't wearing a HR belt that day, but I'm thinking my HR was probably a bit high right then.

Jughed

The only time I ride at VO2 max levels is when I'm doing VO2 max intervals... that number isn't really important for me when it comes to my typical riding style.

Back in my 20's, I had some breathing issues when riding. Insurance was good back then, doc ordered up all kinds of testing, VO2 max was one of the tests. And like Hermes said - there was nothing fun about that test.

That being said, the Garmin estimate is equal +/- to what some online calculators say. 6 min power x weight in KG.

I work fairly hard on the bike and do an OK job with a training routine - my VO2 max is slow to improve, and most of the gains are probably more a function weight loss than fitness gains.

What's really important to me, and probably for most endurance type riders - is the ability to function comfortably at a higher % of VO2 max. Or to not have a section of a climb that puts you into VO2 max power zones completely blow you up during a ride.

...and back in my 20's, my VO2 max was some 17 points higher than it is now - and I wasn't much faster back then over long rides, didn't climb as well as I do now. But I did have a higher short term top end, higher sprint power. And that only really came into play during my Thursday night hammer group ride.

PeteHski

The way I see it is that VO2 max is only a proxy for potential cycling performance or any other aerobic sport. So if you are measuring cycling performance directly i.e. Watts/kg then a backward estimate of VO2 max is not really of much practical interest. My training App uses ride data to estimate W/kg over different time periods and tracks those automatically without any specific test protocols. It reports trends of my power and W/kg for endurance, threshold, max aerobic, anaerobic and sprint, which is far more relevant to cycling specific performance than VO2 max - either estimated or measured directly in a lab.

PeteHski

I think Power/HR is more useful for long steady state efforts like an alpine climb. Not so much for an adrenaline fuelled traffic light hammerfest!

For example I do all my training down at sea level, including specific alpine climb simulations on my indoor trainer. When I ride those climbs for real at altitude I then use my HR to adjust my steady state climbing power. So say I was climbing at a steady 250W with a stabilised HR of 150 on my trainer at sea level, I would use that HR to adjust my power at altitude. So at 2000 m I might end up climbing at say 220W at 150 bpm. I have found this approach very useful on epic alpine events, where pacing is critical to even making it to the finish.

For shorter, high intensity efforts I don't pay much attention to HR as there is way too much lag and all sorts of other variables affecting the HR response. If I'm on a steep punchy climb I will just keep an eye out for when I might start bouncing off my HR rev limiter and back off accordingly if the top of the climb is not in sight. Otherwise I will just ride as hard as I feel I can in the moment.

Iride01 

Quite a while back last year or early this year, I noticed my maxHR on my Garmin Edge was wrong. It'd apparently gotten pushed up to 190 bpm when I had a HR strap fail on me that reported 5 to 10 minute periods of my rides at 190 to 210 bpm. When I adjusted my maxHR down to a more correct number, my VO2 max took a hit soon after. I don't know whether maxHR is used directly in the calculations or not, but apparently it will make a difference. Perhaps VO2 max guesstimation on a Garmin device depends on other things that might use maxHR.

For sure if one is using the VO2 max guesstimate for anything, you should double check all your other data is correct. maxHR, weight and etc.

Hermes

IMO, VO2max as an absolute number is a maximum value requiring an all out effort of brain, neurology, heart, blood, lungs and muscle. What is it good for once one has an accurate value? Getting a pro contract, making an Olympic team or some type of development team run by management that considers VO2max determinate.

Per Coggan, training at VO2max, he suggests 110-120% of FTP for 3-8 minutes, increases VO2max capability. However, he does not say that it increases the VO2max measured ceiling. It may.

I do a lot of training at 110% or more. The more intervals I do at 110-120% the deeper my match book becomes. Having a deep matchbook of 3-5 minute 110% FTP capability is an asset for most events, sportivf, group rides, solo climbing or races. My VO2max absolute max may remain unchanged.

Today, USA Cycling is offering tryouts for the junior development team. The metric used for making the first cut is maximum power for a defined interval on the Wattbike. One of the juniors that I train with at the track, who won match sprint this year at junior nationals, did not generate enough power to make the team. USA cycling is looking at raw talent only. One can debate if this is the right criteria and is the best way to search for new talent. The only thing that matters is what USA cycling coaches think matters.

I feel like I am off topic but I am trying to answer the question are these estimator algorithms accurate? No. And besides that not relevant for most masters and elite athletes' training. But they are fun to play with and discuss on the internet and it is always great to see a a power or Hr metric increase versus go down.