Did your longevity protocol work? Here's how we could find out

Walk into a longevity clinic today and you'll leave with a stack of numbers: a blood panel, a body scan and often a biological age, on top of the sleep and fitness scores from your wearable. Then you start a protocol, maybe an exercise plan, a few supplements or a medication, and a year later you come back to test again.

So, did it work?

It's harder to answer than it sounds, because one person's results can rise or fall for reasons that have nothing to do with the protocol. To really know, you'd need to follow hundreds of patients on it, all measured the same way from the start.

That kind of shared measurement doesn't exist yet, even though clinics have plenty of numbers. When the International Institute of Longevity and the Longevity Think Tank surveyed 22 longevity clinics in 2026, 86% were already using some kind of biological age assessment. Yet the report concluded that outcomes still can't be compared from one clinic to the next.

I care about that gap because Neogenic makes software for longevity clinics and their patients. We've started exploring one possible way to compare outcomes fairly, an idea we call Functional Years.

What your wearable measures, and what it calculates

Start with wearables: watches, rings like Oura and bands like WHOOP. They feel as if they're measuring your fitness, your sleep and your recovery, but what they measure directly is simpler: mostly your pulse, read by light shining into your skin, and the way you move. Everything else is calculated from those signals by each company's own algorithm.

Some of those calculations are good, and some are rough. Here's how recent studies compare them with the lab:

• Sleep on the wrist: in a 2025 study of 62 adults, six wrist-worn devices, including Apple Watch, Fitbit, Garmin and WHOOP, showed only fair to moderate agreement with a full sleep-lab recording on the stages of sleep. The authors wrote that every device still needs improvement.
• Sleep on the finger: in a 2025 study of sleep-clinic patients, the Oura ring matched the sleep lab's stages 53% of the time, and the authors warned that good averages hide large errors for individual people.
• VO2max (how much oxygen your body can use at full effort): a 2025 review of 13 studies, mostly of Garmin watches, found the estimates valid for groups of healthy, untrained adults. They were less reliable for very fit athletes, and some devices overestimated low fitness while underestimating high fitness.
• Apple Watch: a 2025 study found it underestimated VO2max by about 6 points on average, with an average error of 13%. The points are millilitres of oxygen per kilo of body weight per minute, and the authors concluded the watch needs more refinement before clinical use.

Six points may sound small. Doctors often count fitness in METs, and one MET (the energy your body uses just sitting still) equals 3.5 of those points. So the average miss is almost two METs, and large studies link every single MET of fitness to a clear difference in the risk of death.

For a clinic, the bigger problem is consistency. Every brand has its own algorithm, and algorithms change with software updates. Switch devices, or let the company improve its model, and your history can shift while your body stays exactly the same.

None of this makes wearables useless. They're good at showing your own trend, day after day, on the same device. They just aren't precise enough to tell a clinic whether a protocol worked.

Tests of what your body can do

There's an older, less glamorous kind of test that measures what your body can actually do: how fit you are, how strong you are, how fast you walk and how quickly you think. Several of these tests have decades of evidence behind them. A 2024 overview of 26 meta-analyses (studies that pool many earlier studies) called fitness a strong and consistent predictor of illness and death.

• Fitness: across 33 studies and about 100,000 people, each extra MET of fitness, measured with an exercise test, was linked with 13% lower risk of death from any cause.
• Getting fitter: in a study that followed 14,345 men for about 11 years, every MET they gained over time was linked with 15% lower risk of death.
• Strength: in the PURE study of 139,691 people across 17 countries, every 5 kg less grip strength was linked with 16% higher risk of death.
• Walking speed: across nine studies of 34,485 people aged 65 and over, every extra 0.1 metres per second was linked with about 12% lower risk of death.
• Thinking speed: timed tests such as the Digit Symbol Substitution Test, where you match symbols to numbers as fast as you can, are part of the test set in XPRIZE Healthspan.

For clinics, the second study matters most, because it's about change. Men who became fitter over time had lower risk, and change is exactly what a clinic is trying to create. It also puts wearables in perspective, since their VO2max error can span almost two of those steps.

These are associations, and that matters. Fitter, stronger people live longer, but that alone doesn't prove that pushing a number up will add years to your life. What these tests do capture is something you notice every day, like climbing the stairs without stopping or keeping up on a long walk.

Why no biological age test is validated yet

Biological age tests work differently. Most of them run your blood results through a "clock" (a formula that turns those results into an age, built by studying thousands of people of different ages). Some clocks use routine lab values, while the best known, epigenetic clocks, read chemical marks on your DNA that shift as you age. The idea is promising, but the validation isn't there yet.

In May 2026, the Reagan-Udall Foundation, which supports the US Food and Drug Administration, held a meeting on treatments for ageing. Its summary says there's still no FDA path to approve a treatment for ageing itself, and no validated ageing biomarker (a measurable sign of ageing that regulators accept).

The summary also records the FDA's chief scientist as saying that no ageing biomarker has reached validated or reasonably likely surrogate endpoint status.

The research explains the caution:

• A 2026 analysis in Nature Medicine pooled 51 studies of anti-ageing treatments and found that some newer clocks do respond. The authors still say the clocks need validation before a change can be read as better health (Yale News).
• The Scientist reported in 2026 that Steve Horvath, one of the pioneers of epigenetic clocks, said the field still lacks reference standards, and that the lack of standard methods means results for a sample can vary significantly.
• The same report notes that most companies selling biological age kits don't say which clock they use, and that the clocks were designed for studies of large groups, not for judging one person.

So the next time someone tells you a protocol made you three years younger, ask two questions. Which clock? And would another lab give you the same answer?

I'd keep clocks in the picture as research tools, reported beside the main result. It's just too early for them to be the result.

Compared to what?

Even the best test has a problem on its own: it can't tell you whether the care worked, because your body is ageing the whole time. In the Baltimore Longitudinal Study of Aging, VO2max fell by about 3–6% per decade in people's 20s and 30s. By their 70s, it was falling by more than 20% per decade.

So imagine a 72-year-old whose fitness stays exactly the same for a year. On paper, nothing happened. Against what usually happens at that age, it's a real win, and a clinic that only reports improvements would miss it completely.

Regulators are starting to ask for outcomes. Since the second half of 2026, Abu Dhabi's Department of Health has required licensed longevity clinics to report outcome indicators twice a year, including VO2max, blood pressure, LDL cholesterol (the "bad" cholesterol) and HbA1c (your average blood sugar over the past few months). Once clinics report the same numbers, people will compare them, and that comparison has to be fair.

Fair is harder than it sounds, because a few common traps make results look better than they really are. Each one can be reduced, as long as the safeguard is built in from the start:

Trap	What happens	How to reduce it
Testing on a bad day	The next test usually looks better even if nothing was done. This is called regression to the mean.	Take two starting tests and use their average.
Counting only people who come back	Patients who retest are often the ones it's going well for.	Count everyone who started the protocol.
Different devices and instructions	Results shift between clinics for reasons that have nothing to do with care.	Use the same devices and scripted instructions everywhere.
No allowance for age	Clinics with younger patients look better, even when their care is no better.	Compare each result with the change expected for that person's age, sex and starting point.
Chasing the number	Once a number becomes the target, people work on the number instead of the health behind it. Economists call this Goodhart's law, after Charles Goodhart, who described it in 1975.	Show each area beside the total and keep every raw result.

How Functional Years works

Functional Years builds the most important adjustment, the one for age, sex and starting point, into the measure itself. The other safeguards are rules for how clinics collect the data. For every test, the measure asks one question: did this person do better or worse than we'd expect, and by how many years?

1. Estimate where someone of this age, sex and starting level would normally be a year later, using data from large groups of people followed over many years.
2. Compare that expectation with the actual result.
3. Divide the difference by how much function people of that age, sex and starting level normally lose in a year, which gives the answer in years.

Here's the arithmetic for one test, with made-up numbers:

Step	Value
VO2max at the start	30
Expected a year later (a typical drop of 1.5%)	29.55
Actual result a year later	31.2 (4% up)
Gap between actual and expected	5.5 percentage points
Gap divided by the usual yearly loss of 1.5%	About 3.7 years of typical decline avoided

We're proposing four areas, weighted equally to start with. Each one is always shown next to the total, so a loss in one area can't hide behind a gain in another.

Area	Test
Fitness	VO2max from an exercise test
Strength	Chair-stand test (how fast you can stand up from a chair five times) and grip strength
Mobility	Usual walking speed
Thinking speed	A timed thinking test, such as the Digit Symbol Substitution Test

Blood pressure, LDL cholesterol and HbA1c would sit right beside the score. Large trial reviews show that lowering blood pressure and LDL cholesterol lowers the risk of heart attacks and strokes. We'd still keep them outside the score, because they can move a lot with medication alone, and a function score shouldn't reward prescribing.

Biological age clocks and wearable data would sit beside the score too, as supporting signals.

We didn't invent the core idea. XPRIZE Healthspan, a $101 million, seven-year competition for treatments that restore function in people aged 50 to 90, sets its success thresholds as the change equal to 10, 15 or 20 years of expected decline, adjusted for age, sex and starting function. Functional Years explores the same idea in everyday clinic care.

Where we could be wrong

Functional Years has real limits, and some of them we can already see:

• The combined score hasn't been tested against death or disability. Its parts have strong evidence behind them, but the combination is new.
• More exercise training may not always mean a longer life. In the Generation 100 trial in Norway, five years of training in 1,567 adults in their seventies didn't reduce deaths overall compared with the national activity guidelines, though the control group was quite active too.
• In younger, fitter people the normal yearly decline is tiny, so a small gain could turn into a large number of years. We may need to start scoring at 40 or 50, or cap the result.
• Everyday clinic data can leave big questions open. For decades, hormone therapy after menopause was judged by watching women who chose to take it, and it took the Women's Health Initiative trial of 16,608 women, with treatment assigned at random, to show that the combined regimen's overall health risks exceeded its benefits.

Functional Years would compare protocols using exactly this kind of everyday data, so it could be fooled in the same way. That's why I believe we have to plan for being wrong from day one:

1. Before each stage, write down which results would show we're wrong, before we see the data.
2. Keep every raw measurement, so the history can be recalculated if the method changes.
3. Test the system on answers we already know, such as statins (the common cholesterol drugs) lowering LDL, before trusting it with new ones.
4. Ask independent scientists and clinic doctors to challenge it.
5. Publish the methods, including the results that disappoint us.

If we find a crack in the plan, we'll take the feedback and adjust. That only works if we stay open-minded about the idea, rather than fixed on it.

Why this matters to me

In Why I'm building Neogenic, I asked what people could do with more healthy years. I think most of those years are lived through function, through things as simple as getting down on the floor to play with your grandchildren, or still working at a craft you love in your seventies.

Longevity medicine can't yet see which care keeps people fit, strong and sharp for longer, or for whom. Seeing that will take many clinics measuring the same things in the same way.

Functional Years is still a proposal, and I'm sharing it early so the people who practise longevity medicine can shape it. If you run a longevity clinic or work as a longevity doctor, I'd love to hear where you think it breaks. You can reach us through our contact page.