Smart ring 'fitness age' tracks sleep and steps

Expressing cardiovascular risk as an age is a powerful communication device: telling someone their heart looks older than their birthday lands harder than a percentage. Traditionally such estimates need clinical testing, including measured cardiorespiratory fitness. Consumer wearables now generate a fitness age from photoplethysmography-derived heart rate data, which makes the number continuous and passive rather than a one-off in a lab. The unresolved question is whether that convenience comes at the cost of meaning: does a wearable fitness age reflect real lifestyle variation, or is it just a rearrangement of the resting heart rate and estimated VO2 max that feed into it?

This was a retrospective observational cohort of 442 Ultrahuman Ring users across a 12-month window ending February 2026, analysed by the company's own team. The researchers characterised Cardio Age, the ring's cardiorespiratory fitness age estimate, and deliberately separated lifestyle metrics that have no algorithmic link to the score from those that are direct or indirect inputs. Analyses combined cross-sectional Spearman rank correlations, extreme-group comparisons between users with the youngest and oldest estimated cardiovascular ages, and within-person longitudinal trajectories over the year.

The mean Cardio Age gap — estimated age minus chronological age — was -1.84 ± 2.97 years, and 82.6% of participants were rated younger than their actual age. Independent lifestyle measures showed significant but modest associations: sleep efficiency (r = -0.194), REM sleep (r = -0.203), sleep duration (r = -0.200) and daily steps (r = -0.145). A monotonic BMI dose-response appeared, with underweight participants averaging a gap of -3.73 years versus -0.52 for obese participants. Users with the youngest cardiovascular ages slept 37 minutes longer, had 22 more minutes of REM sleep and 1.8% higher sleep efficiency than those with the oldest. Sustained improvers averaged a 3.24-year reduction, alongside resting heart rate falling 0.8 bpm and estimated VO2 max rising 1.3 mL/kg/min.

The result the authors lean on is that variables outside the algorithm still correlate with its output, which is at least consistent with the score capturing genuine physiological and behavioural variation. But the correlations are weak, the within-person improvements track the algorithm's own inputs, and the fact that more than four in five users were rated younger than their age raises questions about calibration in a self-selected, health-engaged group. Nothing here compares Cardio Age with laboratory cardiorespiratory fitness testing or with hard cardiovascular outcomes. The commercial affiliation of every author is an obvious limitation, and independent validation is the missing piece.

A wearable fitness age is best read as a rough, motivational summary rather than a clinical measurement of your heart. That it moved with sleep and step behaviour in this dataset is mildly reassuring, but weak correlations mean plenty of people with good habits will not see the number they expect. The observation that most users scored younger than their chronological age is a hint that such scores may skew flattering in an app-using population. If you track one, watching your own trend over months is more informative than the absolute figure.