Skip to content
Trust Center

Confidence Levels

Every Signal Record and every claim carries a confidence level. It answers one question: how strong is the scientific evidence behind this? Here is exactly what each level means.

What confidence measures

A BioSignal confidence level rates the strength and consistency of the evidence — not how large the benefit is. A modest, well-proven effect can be High confidence; a dramatic but unreplicated claim is not. See why confidence is not effectiveness for the full distinction. The same four-level scale is used everywhere confidence appears on the platform.

High confidence
Definition
Well-supported by consistent, high-quality human evidence.
What it means
The conclusion is unlikely to reverse with the next study. This is about as settled as nutrition science gets.
Typical evidence quality
Multiple randomized human trials and one or more rigorous systematic reviews or meta-analyses, agreeing across independent groups.
Typical number of studies
Many independent human studies, usually synthesized in reviews — not a single result.
How to interpret it
You can act on this with reasonable assurance. Remember: High confidence describes the strength of the evidence, not the size of the effect.
Moderate confidence
Definition
Promising evidence that is still developing or context-dependent.
What it means
The finding is more likely true than not, but important gaps or inconsistencies remain.
Typical evidence quality
Some human trials support it, but they may be smaller, shorter, mixed in result, or limited to specific populations.
Typical number of studies
A handful of human studies pointing the same way, without the depth or consistency of a High rating.
How to interpret it
Reasonable to consider, with the understanding that future research could strengthen or temper the conclusion.
Limited evidence
Definition
Early or sparse human evidence; conclusions remain uncertain.
What it means
There are signals worth noting, but not enough reviewed human evidence to be confident either way.
Typical evidence quality
Mostly mechanistic, preliminary, indirect, or animal evidence, with little or no strong human trial data.
Typical number of studies
Few human studies, often small or preliminary — or the case rests mainly on mechanism and preclinical work.
How to interpret it
Treat as an open question, not a recommendation. Interesting, but unproven in people.
Caution
Definition
Conflicting evidence or known risks that warrant care.
What it means
There is a specific reason for care — either credible risk information or evidence that meaningfully conflicts.
Typical evidence quality
Documented safety concerns, contraindications, or well-conducted studies that contradict the popular claim.
Typical number of studies
Variable — the flag is driven by the nature of the evidence (risk or conflict), not the count of studies.
How to interpret it
Slow down and read the safety section. This is where independent, transparent review matters most.

A note on reading the score

Confidence can move in either direction over time. As new trials are published, a Limited rating can rise to Moderate, or a Moderate finding can be tempered. That is the system working as intended — every such change is recorded in the record’s version history.

Spotted something unclear or out of date? We read every note.

Give feedback