digital biomarker

A digital biomarker is a number your smartwatch or phone tracks every day—like heart rate or sleep duration—that has actually been verified to relate to a specific disease or health condition. Think of it like a thermometer: a temperature reading by itself is just a number, but it becomes a meaningful signal once you attach a rule, such as "above 38°C suggests infection." In the same way, out of the massive pile of numbers wearables produce, only the patterns that have been tested and shown to genuinely correlate with something like depression or diabetes earn the name digital biomarker.

The problem is that wearable data is so vast that it's hard for humans to manually pick out which patterns actually matter. Combinations that aren't obvious at a glance—like how erratic nightly sleep duration is, or the ratio between step count and resting heart rate—can turn out to correlate with real health conditions. That's why AI is increasingly being used not just to spot candidate indicators within huge streams of biosignals, but also to statistically verify whether those indicators are genuine correlations rather than coincidences.

In articles, it's used in phrases like "a multi-agent system that discovers candidate digital biomarkers from wearable data." A common misunderstanding is that identifying a digital biomarker doesn't mean an immediate diagnosis. It's just a reference indicator with confirmed statistical correlation—actual diagnosis is still up to medical professionals.