The Hidden Science on Your Wrist: How Smartwatches Measure Health

Smartwatches and fitness bands promise a lot in a small package: heart rate, sleep, stress, oxygen saturation, and even heart-rhythm alerts. The key idea is simple. Most wearables do not “see” disease directly. They measure physical signals at the skin, then use algorithms to estimate what is happening inside the body. That gap between signal and interpretation is where both the usefulness and the limits of wearables come from.

The most common sensor in a smartwatch is photoplethysmography (PPG). PPG uses a light source and a photodetector on the skin to measure tiny changes in light absorption or reflection as blood pulses through vessels. A widely cited review explains PPG as a non-invasive optical method that tracks volumetric changes in blood circulation, which is why it can be used to estimate heart rate.

In practice, many wearables shine green light into the skin because it is strongly affected by blood flow near the surface. The device then detects repeating peaks in the reflected signal and converts that pattern into beats per minute. This is why heart-rate readings are often accurate when you are still, but become less reliable when the signal is noisy.

Some wearables also estimate SpO₂ (oxygen saturation) using a pulse-oximetry style approach. Traditional pulse oximetry relies on red and infrared light absorption differences to estimate oxygenation. However, this kind of measurement is sensitive to real-world conditions. International standards for pulse oximeter equipment explicitly discuss accuracy requirements and the need to disclose performance under conditions such as low perfusion if manufacturers claim accuracy there.

Independent clinical literature also highlights that pulse oximetry accuracy can vary, including concerns about skin pigmentation and measurement bias in some contexts. For example, an article in the British Journal of Anaesthesia discusses evidence of inaccuracy with darker skin pigmentation and the importance of understanding how pulse oximetry works to appreciate limitations. This does not mean SpO₂ features are useless. It means you should treat them as estimates that can be affected by the situation, not as perfect medical readings.

When a smartwatch offers an ECG feature, it is not using light. It is using electrodes to measure the heart’s electrical signals, usually as a single-lead ECG. This is closer to clinical testing than PPG is, but it still has strict limits.

At the same time, many watches also offer rhythm alerts that are not ECG readings. For example, Apple’s Irregular Rhythm Notification is a background screening feature that uses pulse-rate signals collected by the Apple Watch’s PPG sensor to look for patterns that may be consistent with atrial fibrillation. This is different from the Apple Watch ECG app, which records an electrical waveform using electrodes. In other words, even when a feature sounds “medical-like,” what the device is actually measuring matters, and so does the intended use.

Wearable measurements are vulnerable because the body is not a controlled lab environment. A few factors cause the biggest errors.

Motion is a major one. When you run or move your wrist, the watch shifts and the optical signal changes for reasons unrelated to blood flow. Low perfusion is another issue. Cold skin or poor circulation reduces the pulsatile signal that PPG and pulse oximetry depend on, and standards for pulse oximetry performance specifically address how accuracy claims relate to low perfusion conditions.

Fit and placement matter too. A loose strap lets light leak in and makes the signal unstable. Tattoos, sweat, and ambient light can also interfere. Finally, human variation matters. Optical sensing interacts with skin and tissue properties, which is why researchers discuss limitations such as pigmentation-related effects for pulse oximetry and the need for careful interpretation.

Wearables are powerful because they make invisible signals visible. PPG can estimate heart rate by tracking blood-volume changes through light, and related optical methods can estimate oxygen saturation under good conditions. However, real-world factors such as motion, low perfusion, fit, and individual differences can degrade accuracy, and some features are explicitly described as screening tools rather than diagnostic systems. The best way to use a smartwatch is to treat it as a tool for patterns and early signals, while relying on medical testing when decisions truly matter.