How accurately does my smartwatch measure heart rate?

At rest, optical heart rate measurement is quite reliable: reviews find a mean deviation of around ±3 percent against the ECG. During movement, with a loose fit, tattoos or cold skin, however, the error rises markedly. A device is not sufficient for a medical assessment of heart rhythm.

Can I rely on the sleep-stage display?

Only to a limited extent. Devices detect well whether you are asleep or awake but struggle with brief periods of wakefulness and the precise distinction between light, deep and REM sleep. In validation studies against the sleep laboratory, agreement on individual stages was only fair to moderate. The values are useful for rough trends, not for a precise sleep analysis.

Is my wearable's SpO2 value medically usable?

As a rule, no. The SpO2 function of many consumer devices is explicitly not an approved medical device. Moreover, optical sensors systematically overestimate oxygen saturation on darkly pigmented skin, so that a genuine deficiency can be overlooked. If respiratory or cardiovascular problems are suspected, a medical evaluation is necessary.

Longevity Tracking7 min readLast reviewed: June 2026

Wearables in Longevity: What Trackers Really Measure

Smartwatches and fitness rings have become a fixture of the longevity scene: they promise to make health, recovery and the ageing process measurable. They do indeed deliver continuous data on heart rate, heart rate variability (HRV), sleep stages and oxygen saturation (SpO2) – values that used to be obtainable only in a laboratory. But between what a device displays and what it actually measures reliably there is often a considerable gap. This article explains, in an educational way and without hype, which metrics wearables capture, how accurate research considers them to be and where the line runs between meaningful self-observation and data overload. PeptidLotse does not provide medical advice here – the assessments that follow are no substitute for a medical evaluation.

Machine-assisted translation. The German original is the authoritative version.

Key points

Wearables measure most values indirectly via optical sensors and algorithms – they approximate clinical references but do not replace them.
The most accurate metric is resting heart rate (≈±3 %); HRV is only reliable at rest, and sleep stages are captured only roughly.
SpO2 measurements are prone to interference and systematically overestimate the value on dark skin – consumer functions are usually not a medical device.
Individual trends over time are more meaningful than absolute values from single nights; software updates can change measurements unnoticed.
There is no evidence that wearing a tracker extends lifespan; conspicuous values belong in the hands of a doctor.

What wearables measure – and how they do it

Most devices worn on the wrist or finger do not capture values directly but derive them from indirect signals. Central to this is photoplethysmography (PPG): an LED shines light into the tissue and a sensor measures how much light is reflected by the pulsating volume of blood. From this signal the software calculates heart rate, estimates heart rate variability (the fluctuation in the intervals between heartbeats) and – via light absorption at different wavelengths – oxygen saturation. Sleep stages, in turn, are not measured directly but estimated from movement (actigraphy), heart rate and HRV patterns using algorithms.

The underlying principle matters: a medical ECG measures the heart's electrical activity directly, and a clinical sleep laboratory (polysomnography) records brain waves, eye and muscle movement. Wearables approximate these reference values through workarounds and modelling assumptions. That explains why the same night can look different on two devices.

PPG (optical pulse sensor) is the basis for heart rate, HRV and SpO2
Sleep stages are estimated algorithmically, not measured directly
The reference methods are the ECG (heart) and polysomnography (sleep)
Different manufacturers use different, mostly undisclosed algorithms

What the research actually shows

A living umbrella review in Sports Medicine (2024) summarised 24 systematic reviews comprising 249 validation studies and around 430,000 participants. For resting heart rate, wearables performed well: the mean deviation was about minus 3.4 beats per minute, that is roughly ±3 percent. Step count and energy expenditure were considerably less accurate, and sleep tended to be overestimated in duration, with errors typically above 10 percent. Notably, only about 11 percent of the more than 300 devices examined were validated for even a single metric.

For HRV, a systematic review (Folia Medica, 2018) shows that agreement with the ECG at rest is very good to excellent – but decreases progressively as exertion rises. HRV values at the wrist are therefore most meaningfully interpretable at night or in a resting state. For sleep, a validation study in SLEEP Advances (2025) tested six devices against polysomnography: all reliably detected whether someone was asleep (sensitivity above 90 percent) but struggled to identify periods of wakefulness (specificity 29–52 percent). Agreement on classifying individual sleep stages was only "fair to moderate". Devices are thus suited to rough trends, not to the precise mapping of sleep architecture.

Resting heart rate: high accuracy (≈±3 %)
HRV: good at rest, unreliable during movement/exertion
Sleep: detects sleep vs. wake well, sleep stages only roughly
Step count and calorie expenditure: marked error margins

Limits and systematic sources of error

Optical sensors are susceptible to interference: movement, a loose fit, tattoos, cold fingers and sweat all degrade the signal. Oxygen saturation is particularly relevant. Classic clinical pulse oximeters – and thus also the sensors in wearables – systematically overestimate SpO2 on darkly pigmented skin. The US authority FDA requires an accuracy within about 3 percent against arterial blood gas measurement; real-world data show, however, that so-called occult hypoxaemia (a genuine oxygen deficiency despite a normal displayed value) was overlooked considerably more often in patients with dark skin. It must be stressed: the SpO2 function of consumer wearables is generally explicitly not a medical device and is not intended for diagnosis.

There is also a methodological problem: manufacturers continually adjust their algorithms via software updates without disclosing the changes. A version validated today may measure differently after an update. Absolute values should therefore be viewed with caution – individual trends over time on the same device and under the same conditions are more meaningful.

Movement, skin tone, tattoos and cold disturb the optical signal
SpO2 is systematically overestimated on dark skin (occult hypoxaemia)
Consumer SpO2 is usually not an approved medical device
Software updates change algorithms unnoticed – absolute values are unstable

Data hype vs. meaningful use

In the longevity community, wearable data are often presented as a precise mirror of "biological age" or daily "recovery". Yet such recovery or readiness scores are proprietary composite values drawn from several estimates – not a clinically validated health index. It is a plausible but unproven claim that daily optimisation of such scores extends lifespan. Robust long-term studies showing that wearing a tracker increases life expectancy are lacking.

Used sensibly, wearables are a tool for self-observation: they can make patterns visible (such as how alcohol, stress or late exercise shift resting heart rate and HRV) and motivate physical activity. It becomes problematic when users attach too much importance to individual nightly values, fall into data anxiety or self-diagnose symptoms based on a smartwatch reading. Conspicuous or persistent changes – for instance in heart rhythm, oxygen saturation or sleep – belong in the hands of a doctor and should not be interpreted via app values.

Readiness/recovery scores are proprietary, not clinically validated
No evidence that tracker use itself extends lifespan
The strength lies in trends and self-observation, not single values
For conspicuous values, seek medical evaluation rather than self-diagnosis

Frequently asked questions

How accurately does my smartwatch measure heart rate?: At rest, optical heart rate measurement is quite reliable: reviews find a mean deviation of around ±3 percent against the ECG. During movement, with a loose fit, tattoos or cold skin, however, the error rises markedly. A device is not sufficient for a medical assessment of heart rhythm.
Can I rely on the sleep-stage display?: Only to a limited extent. Devices detect well whether you are asleep or awake but struggle with brief periods of wakefulness and the precise distinction between light, deep and REM sleep. In validation studies against the sleep laboratory, agreement on individual stages was only fair to moderate. The values are useful for rough trends, not for a precise sleep analysis.
Is my wearable's SpO2 value medically usable?: As a rule, no. The SpO2 function of many consumer devices is explicitly not an approved medical device. Moreover, optical sensors systematically overestimate oxygen saturation on darkly pigmented skin, so that a genuine deficiency can be overlooked. If respiratory or cardiovascular problems are suspected, a medical evaluation is necessary.

Sources

This article is for information and education only. It does not replace medical advice and deliberately contains no dosing, usage or sourcing information.

← All guide articles