How Smartwatches Record Biometrics

This document provides a brief overview of the methods smartwatches use to measure and record biometrics, with a focus on Garmin watches.

The watches have three basic sources of raw data, from which all biometrics are derived: an optical technology called photoplethysmography, or PPG (the light-emitting sensor on the back of the watch); internal motion sensors (accelerometers and gyroscopes); and barometric altimeters (built-in barometers that measure changes in atmospheric pressure).

Biometrics recorded by smartwatches fall into two categories: continuous or cumulative. Continuous metrics on Garmin watches include inter-beat intervals, heart rate, pulse-oxygen level, stress, respiration, and body battery. Cumulative metrics include calories burned, floors climbed, intensity minutes, and steps taken — these reset to zero at the start of each day and accumulate as the day goes on.

The only continuous metric reported in real time is the inter-beat interval (IBI). Heart rate is derived from IBI (60000/IBI), but computed as an average across several to many IBI samples rather than from any single one. Every other continuous metric depends on IBI and other factors, and is reported on a delay, once enough validated supporting data has accumulated.

Cumulative data get reported asynchronously and repeatedly, whether or not the value has actually changed. Watches will often record and report motion-dependent biometrics even when they aren't being worn.

When we generate reports that include biometrics, the continuous data shown reflects whatever the watch reported at that specific time. Cumulative data shown reflects the most recently reported value for that day, regardless of when it was reported.

In other words, once a cumulative data element has been recorded, it'll show up in the report even if it hasn't changed in hours. Further, several continuous data elements may appear even after the watch has been taken off the wrist, because all their contributing components — including motion-based ones — have only just been validated.

Finally, when the watch is first put on, motion-dependent data elements are generally reported before IBI and heart rate, since it takes time for the watch to settle against the wrist and its readings to be considered valid.

Smartwatches measure inter-beat intervals — the time difference between successive heartbeats — primarily through an optical technology called photoplethysmography (PPG). When processing biometrics collected from smartwatches, people use a range of different acronyms that, in practice, all refer to the same thing: a cleaned-up version of the intervals between successive R-wave peaks.

How Optical PPG Works

Green or multi-wavelength LED lights on the back of the watch shine light through the skin. As your heart beats, blood volume in your wrist increases and absorbs more light; when your heart relaxes, less light is absorbed. Photodiodes (light sensors) measure the amount of light reflected back from your blood vessels. Software processes that data to pinpoint each peak of blood flow, then calculates the time interval between successive beats — often called pulse-to-pulse or PPG-based RR intervals.

Electrical ECG Method (Alternative)

Some advanced smartwatches feature built-in electrical sensors. Touch a designated spot — the crown or bezel — with your opposite hand, and it completes a closed circuit with electrodes on the back of the watch. This captures a single-lead electrocardiogram (ECG), measuring the heart's actual electrical impulses rather than physical blood flow.

Smartwatches measure blood oxygen by shining red and infrared light through the skin on your wrist and measuring how much light reflects back. LEDs on the back of the watch shine colored and infrared light into the blood vessels of your wrist. Oxygen-rich blood absorbs more infrared light and reflects more red light, while oxygen-poor blood does the opposite. Photodiodes (sensors that turn light into electrical signals) measure the amount of light that bounces back. Embedded software processes these light measurements to estimate your blood oxygen saturation percentage (SpO₂).

Normal resting readings typically range from 95% to 100%, but these are percentages of an absolute value that doesn't account for variation in skin tone, loose bands, or movement. Pulse oximeter readings can easily be distorted by physical, physiological, or environmental factors — several common ones interfere with how light passes through the skin, leading to false highs or false lows.

The benefit of wrist-based SpO₂ measurements can be substantially increased by establishing an n-minute moving baseline against which newly reported values can be compared. Doing so eliminates the need to compare values collected by the watch with those collected in a lab, and mitigates the challenges variations in skin tone otherwise introduce. As long as values are measured consistently, the results are useful.

Body Battery is a daily energy score, from 5 to 100, estimating your available physical and mental reserve energy. It draws on HRV, sleep quality (good sleep — especially deep and REM sleep — increases your score overnight), and activity level (exertion and exercise consume energy based on workout intensity, duration, and your personal fitness level).

The challenge with this metric is that it assumes lower HRV always signals higher stress, and drains your battery accordingly even while you're sitting still. It doesn't account for the fact that reduced HRV can just as easily reflect a period of enjoyment.

A smartwatch computes your respiration rate by tracking tiny, rhythmic changes in your heart rate or blood flow caused by the movement of your lungs.

When you inhale and exhale, changes in chest pressure alter how blood flows back to your heart. This creates a small, rhythmic wave in the light signal recorded by the PPG — a baseline modulation that matches your breathing cycle.

Your heart beats slightly faster when you breathe in and slightly slower when you breathe out — a process called respiratory sinus arrhythmia. Respiration rate calculations aren't immediate; they require long sequences of heart rate data to find the regular, low-frequency pattern that spans hundreds of heartbeats.

A smartwatch computes floors climbed by combining changes in altitude with physical step movements. These changes are primarily recorded by built-in barometric altimeters that track changes in atmospheric pressure, backed up by GPS data for dual accuracy.

Air pressure drops as you climb higher and rises as you go lower. Watches use internal barometers to measure these changes and translate them into vertical feet or meters. Ideally, this is combined with GPS data calculating your position in 3D space, including height above sea level.

The watch counts one floor for every 10 feet (3 meters) of vertical ascent — about 16 steps. Since GPS vertical data can be slow to update during quick climbs, reported changes are often delayed. The floors-climbed metric requires no heart rate or related data.

A smartwatch uses built-in motion sensors — accelerometers and gyroscopes — to measure wrist movement, which is generally chaotic and full of noise (typing, cooking, waving). But when you walk or run, your arms swing in a steady, rhythmic pattern, and that swing correlates with the steps you're taking.

The software sets a threshold for acceleration spikes: if a motion pattern matches the cadence, timing, and amplitude of a human stride, it counts as a step. Random, isolated movements that don't fit that repeating pattern get discarded.

Arm-swing data is reinforced by a second signal: a regular acceleration spike each time your foot hits the ground, sending a subtle shockwave and vertical jolt up to the wrist. That signal helps confirm the same stride pattern the arm swing already suggested.

Because the watch relies on arm swing as a proxy for steps, it misses steps when your arms aren't swinging — pushing a cart or a lawn mower, for instance. Many modern watches add a gyroscope to measure rotation and orientation, which helps confirm that a movement is part of a natural walking gait rather than random arm motion, like reaching for something or checking the time.

Some devices also layer in GPS data to track actual distance and speed, isolating real walking or running from stationary movement. That works well — unless you're running on a treadmill.

Smartwatches compute calories burned by combining data that you specify — age, sex, height, weight — with real-time sensor measurements. From what you provide, the watch calculates the calories your body burns just to stay alive: breathing, circulating blood, keeping organs running. Accelerometers measure the speed, direction, and rhythm of your arm movements to calculate activity level and step count. Optical sensors provide correlated heart rate data to gauge exertion. GPS and barometric altimeters add speed, distance, and elevation change.

All of this feeds into the vendor's proprietary algorithms, which estimate — really, guess — your total active energy expenditure.

Intensity minutes measure time spent in vigorous activity. Smartwatches (Garmin included) compute them by comparing your real-time heart rate to your average resting heart rate, or by tracking your step cadence. Elevated heart rates that fall into moderate or vigorous exercise zones earn credit.

Even without heart rate data, watches can estimate moderate intensity minutes just by counting steps per minute (cadence) — though they can't distinguish vigorous minutes without simultaneous heart rate data.

What smartwatches typically report as "stress" is actually a measure of arousal level. The measurement is based on the variance in time between successive heartbeats — heart rate variability, or HRV. As arousal increases, the variability between successive beats decreases.

The metric carries no emotional valence. Running to catch a bus, feeling stage fright before a performance, savoring a great cup of coffee, working happily on a favorite task, or running into a good friend you haven't seen in years — any of these can spike your arousal level, whether you're stressed by the moment or genuinely engaged by it.

As a result, watches often report high stress even when you actually feel happy or focused. The watch can't tell negative worry from positive excitement.