This series started with my recovery from surgery and one goal: finish the Michigan 70.3 Half Ironman.
I did. It took me just over seven hours, which was an improvement for me and something I’m proud of, even though it wasn’t a competitive time by any stretch. For perspective, the fastest professional men finish a 70.3 in under four hours, and the 2026 men’s world champion went 3:50:51. At this year’s Michigan race, the winner of my 45–49 age group came in at 4:32:31. Depending on the course, a highly competitive age-grouper in my division might aim for somewhere around 4½ to 5 hours.
I wasn’t trying to win my age group. I was recovering from a major illness, rebuilding my fitness, and getting ready for a hard endurance event while working full time and trying to be present for my family. The finish line gave me a concrete endpoint, but the more interesting part was everything I learned getting there.
Over several months, I tracked my training with wearables, cycling performance tests, lab testing, and proteomics. Sleep, nutrition, fatigue, and the type of training I chose all moved the numbers more than I expected. And different measurements sometimes seemed to disagree with each other, because they were capturing different parts of fitness and physiology.
Now that the race is behind me and I finally have a directly measured VO₂ max from a CPET, this last post brings the whole experiment together: what each measurement got right, where the results diverged, and what any of it was actually useful for.
What Is My VO₂ Max? The Lab Finally Answers
I started this series with a simple question: can the measurements available outside a physiology lab meaningfully describe human fitness?
I ended up with more answers than I bargained for. My Apple Watch and WHOOP estimated VO₂ max from wearable data. VO Health estimated it from proteins in my blood. A Peloton ramp test predicted it from peak cycling power. Functional threshold power (FTP) and critical-power time trials gave me still more estimates, based on what I could produce and sustain on the bike.
The results ranged from the low 40s to the low 60s.
I ultimately completed the gold standard test to help me compare them all: a cardiopulmonary exercise test, or CPET, with direct respiratory gas analysis.
My measured VO₂ max was 50.1 mL/kg/min, right in the middle of all the estimates. But the more interesting result, to me, was how the other measurements scattered around that number and how they changed over time.
The Final Comparison
These tests weren’t all done on the same day, so this is a personal longitudinal experiment, not a formal validation study. Still, lining them up shows how differently each method interpreted the same stretch of training.
| Measurement | Initial result | Follow-up result | Difference from CPET |
| Apple Watch | 49.3 | 47.0 | -3.1 |
| WHOOP | 50.0 | 48.0 | -2.1 |
| Molecular VO₂ max | 54.2 | 49.1 | -1.0 |
| Peloton ramp-test estimate | 61.7 | 62.6 | +12.5 |
| FTP-derived estimate | 41.7 | 43.2 | -6.9 |
| Five-minute time-trial estimate | 42.9 | 44.4 | -5.7 |
| Direct CPET | — | 50.1 | Reference |
The closest was the Molecular Fitness Test from VO Health, which estimated my VO₂ max at 49.1 a few weeks before the CPET measured 50.1. WHOOP and Apple were also reasonably close, and the cycling calculations missed from opposite directions.
The molecular result stood out because it required no exercise at all. It came from a blood sample, not from heart rate, pace, power, or expired gas. That doesn’t prove a blood test is the most accurate way to estimate VO₂ max. This is one comparison in one person, the tests were weeks apart, and a close match can happen partly by chance. Still, landing within 1 mL/kg/min of my measured value is an encouraging data point, and research has consistently shown that circulating protein patterns can capture biologically meaningful features of cardiorespiratory fitness and can change with exercise training.
A note on the molecular numbers: the table shows the VO Health results as originally reported. During my training, VO Health updated its model, and my results changed with it. Under the updated model, my three tests came out at 50.9, 50.8, and 49.1, all within about two points of my CPET result. The final report already used the updated model, which is why 49.1 is the same either way. I’ve kept the original numbers in the table because the updated model came later. VO Health keeps refining its model as it collects more calibration data in an effort to continue to refine their accuracy.
Why Did My Numbers Move in Opposite Directions?
What I found most useful wasn’t which estimate came closest to 50.1. It was what happened during training.
My cycling got better:
- Ramp-test peak power went from 330 to 335 watts.
- FTP went from 168 to 177 watts.
- Ten-minute power went from 181 to 195 watts.
- Five-minute power went from 202 to 213 watts.
- Two-minute power went from 230 to 245 watts.
At the same time, Apple’s estimate dropped from 49.3 to 47, WHOOP’s dropped from 50 to 48, and the molecular estimate fell from 54.2 to 49.1 (using the original model; see note above. After the update, the decline was less, 50.9 to 49.1).
At first glance that looks contradictory. It makes more sense once you ask what question each test is really answering.
I spent a lot of time on the bike, repeatedly practicing the same kinds of efforts the performance tests use. That probably improved my cycling economy, pacing, muscular endurance, tolerance for discomfort, and ability to sustain a bigger fraction of my aerobic capacity. Those are real gains, but they’re not the same thing as an increase in maximal oxygen uptake.
The ramp test also reflects more than aerobic capacity. Peak power is influenced by anaerobic contribution, cadence, motivation, and how familiar you are with the protocol. That likely helps explain why its VO₂ max estimate stayed so much higher than my CPET result.
FTP and critical power answer a more practical cycling question: how much power can I sustain? For an Ironman bike leg, that may be more useful than knowing my maximal oxygen uptake. But the equations that convert those numbers into VO₂ max are population-based estimates, not direct measurements of my own physiology.
The wearables were looking at a different version of my training.
Apple estimates cardio fitness mainly from qualifying outdoor walks, runs, and hikes. A big share of my training was indoor cycling and swimming, which contributed little or nothing to that estimate. I was also running less while getting better on the bike, so Apple could report a decline even as my cycling improved.
WHOOP uses a broader proprietary model, but it still depends on the relationship between heart rate, activity, recovery, and performance. Built-up fatigue, poor sleep, heat, dehydration, stress, and time away from running can all shift that relationship without meaning I’d actually lost maximal aerobic capacity.
It’s also likely that the small decline is real. Overtraining without enough recovery, or training in the wrong way, can pull maximal aerobic capacity down, and as an amateur without a coach I’m easily guilty of both. I leaned heavily on the bike, ran too little, and often trained on too little sleep, so I can’t rule out that my underlying aerobic fitness slipped a bit even while my cycling-specific numbers improved. With only one direct CPET measurement, I can’t say for sure. That’s one more reason I’d rather treat the estimates as a trend to investigate than a verdict.
The Conditions Around the Test Matter
Sleep and nutrition became impossible to ignore during this experiment. Sleep deprivation measurably hurts endurance performance and raises perceived effort. Carbohydrate availability and hydration affect time-trial performance and your ability to keep going over long efforts. Getting these right matters most for performance-based tests, though it shouldn’t be confused with big overnight changes in true VO₂ max.
For a planned test or race day, I could prepare. I could cut back my training load, prioritize sleep, hydrate, eat well, and show up ready to give a maximal effort. I organized my life around that one day to give the test its best shot at capturing what I could do under good conditions.
The months of training before the race looked very different. My sleep and nutrition were often far from ideal. I regularly stayed up late finishing leftover work and then got up early to train. Workouts had to fit around a full-time job and family responsibilities. Meals weren’t always timed well, recovery was often incomplete, and the next session still had to happen.
That may partly explain why my cycling performance improved while my Apple and WHOOP estimates fell. A carefully prepared test shows what I can do under near-ideal conditions. My wearable data captured the cumulative effect of ordinary life, including the many days I trained on too little sleep, carried fatigue from the day before, or didn’t fuel well enough.
Why More Than One Measurement Helps
CPET is the best direct measurement of maximal aerobic capacity. The cycling tests showed whether I was getting better at producing and sustaining power on the bike. Apple and WHOOP showed how heart rate and activity interacted across weeks of real-world training. The molecular panel gave me an exercise-independent estimate, plus some information about the biological pathways associated with fitness.
No single number can capture all of that.
The value was in seeing where the measurements agreed and where they pulled apart. If VO₂ max, power, pace, and submaximal heart rate all improve together, the story is pretty straightforward. If cycling power goes up while wearable estimates go down, it raises questions:
Am I getting more efficient on the bike without increasing my maximal aerobic capacity? Am I carrying too much fatigue? Have I neglected running? Is my sleep limiting recovery? Is a testing protocol rewarding familiarity more than actual physiological change?
When the numbers disagree, that’s information you can act on.
That doesn’t mean everyone needs six ways to estimate VO₂ max. It means the measurement should fit the question you’re asking. Someone training for a cycling event may care most about FTP and critical power. Someone tracking their health over years may get more from a consistent wearable trend. Someone who can’t, or doesn’t have the time to, do maximal exercise testing might benefit from an estimate that doesn’t require exercise at all.
Fitness as a Clinical Measurement
VO₂ max matters well beyond amateur endurance sports.
Cardiorespiratory fitness is strongly associated with cardiovascular disease, functional capacity, and mortality. The American Heart Association has argued that it should be treated as a clinical vital sign because it adds prognostic information beyond many traditional risk factors. Yet we rarely measure it in routine care.
A formal CPET takes specialized equipment, trained staff, time, and a maximal effort. That’s appropriate when you need detailed physiology, but it doesn’t scale as a routine screening test for most healthy adults.
Wearables have made estimated fitness far more accessible, though accuracy varies from person to person. A blood-based estimate could lower the barrier even further if it proves accurate, reproducible, responsive to meaningful change, and valid across diverse populations – VO Health is delivering on this.
I can imagine cardiorespiratory fitness eventually sitting alongside blood pressure, cholesterol, and body composition as a familiar health measurement. It wouldn’t replace CPET when direct physiological assessment is needed, but it could give people a baseline, flag an unfavorable trend, and start a conversation about exercise, recovery, and long-term health.
Summarizing What I Learned
My CPET result was 50.1.
The Molecular Fitness Test came closest at 49.1, with WHOOP and Apple not far behind. The cycling estimates were scattered much more widely, but they were still valuable because they documented something that mattered to my trajectory: I was becoming more efficient on the bike, which is where I’d put most of my training.
That sums up the main lesson of this series. VO₂ max, wearable fitness, molecular fitness, and sport-specific performance overlap, but they aren’t interchangeable. Figure out which metrics matter most for your question, focus on those or combine them, and adjust as you go. The goal isn’t to force them all into agreement. It’s to understand what each one sees, follow the trends, and dig in when the story changes.
For me, the numbers ultimately converged on a VO₂ max of about 50. They also exposed the limits of my training: strong short-duration cycling, improved sustainable power on the bike, inconsistent recovery with lack of sleep, and the predictable strain of preparing for an endurance race around a full life.
That’s a lot more useful than a single score.
Earlier Posts in This Series
- Beyond the Numbers: Can We Actually Measure Human Performance?
- VO₂ Max: Beyond the Laboratory
- Molecular Fitness: Where the Blood Test Lines Up
Disclosure: VO Health supported this project and provided the Molecular Fitness testing discussed in this series. I evaluated its results alongside my wearables, performance tests, and direct CPET. I’m grateful for their support and for helping me think more carefully about what these measurements can tell us.
