What actually works for you?

Every wearable hands you a recovery score. None of them tell you which of the things you are doing is working, for you, on your own measurements. Health answers that, and shows its working.

Live now Open Health See where you're heading

01A score is not an answer

every wearable

Your recovery is 64%. Readiness is down.

A number about you, built from a population model you were never fitted to. It cannot separate your signal from your noise, so it cannot tell you whether the late session, the alcohol or the travel is the thing that matters.

health · your own effect

Evening training is costing you 3.9 bpm.

Between 2.9 and 5.0, on your own nights, at your own measurement error. That is a decision you can act on tonight, and when the evidence does not support one we say so instead of estimating anyway.

your squad

Your squad is what makes it fast.

Alone, an answer takes the best part of a year. With a squad measuring alongside you it takes about twenty-eight nights.

tested on real people

13 out of 13 got a sharper number.

Across 1,623 nights of real measurement, every person ended up with a threshold around five times tighter than the device figure: 2.5 beats against 13. And on a trial where the drug genuinely did nothing, we told all 38 patients it did nothing.

02How to start

Day 1

Connect the ring or watch you already wear. No new hardware, nothing to log.

now

Day 8

Eight nights of sleep and we know your own normal range. Everyone in our testing had theirs by day eight.

now

Day 12

Change one thing, some days on, some off, and get the size of the effect on you and how sure we are, or we tell you it is too small to bother with.

now

03Where you're heading

Your next four weeks, before they happen.

Resting heart rate, HRV and jump height, projected four weeks ahead from your own history, with the odds that you move outside your own normal range, week by week.

Your readings are opened with your passkey and forecast on your phone or laptop. Nothing is sent to us. Once five teammates share their summaries, the forecast is tuned to your squad, using a handful of numbers from each of you, never the readings themselves.

More accurate than a leading AI forecasting model when both were tested on the same 233 people from four public research datasets, none of whom were used to build it.

A pulse traced in dots inside a pale green band bounded by two green lines, with faint dotted lines above and below.

04The instruments you already own

2.79 cm

Countermovement jump on a force plate. ICC 0.93, CV 2.5%. A wearable sensor needs 8.63 cm on the identical jump.

now

166–415

VO2max by true CPET, in mL/min, against ±12 mL/kg/min for a wearable estimate.

now

5.3%

Bone density, lumbar spine: the ISCD least significant change, set by your own facility.

now

≥15%

Eccentric hamstring peak torque, held to the published floor.

now

150 mL

Spirometry FEV1 and FVC: the ATS/ERS within-session repeatability criterion.

now

05What we refuse

Refuted

Your DNA does not predict how you respond to a diet. A randomised trial at the extremes of genetic predisposition: P = 0.94.

refuted

Refused

Limb asymmetry as an individual verdict. Bilateral strength ratios reach ICC below 0.60.

refused

Refused

Ranking teammates against each other.

refused

Refused

Per-food verdicts from one CGM pass. Around 90% of individual food effects fall inside the within-person noise of a single pass.

refused

Trend only

Gait speed and heart-rate variability: shown as a trend, never as a verdict.

trend

06Locked before we look

The analysis is pre-registered.

Every threshold and decision rule was fixed in advance and rehearsed end to end, so the cohort cannot be read into telling us what we hoped.

Figures from measured results on our own system and from the published literature. Device errors against an ECG criterion; DXA thresholds per ISCD Official Positions; spirometry per ATS/ERS. Validation figures from PMData (doi 10.1145/3339825.3394926) and the Raynaud n-of-1 series (NCT02050360), both openly licensed.

What actually works for you?

Your next four weeks, before they happen.

The analysis is pre-registered.

Measured, not predicted.