Methodology

Evidence tiers

Evidence is shown on a visible seven-tier scale (1 = human lifespan/clinical-outcome RCT, down to 7 = in vitro/ex vivo) and is never collapsed into a single marketing score. Tier 1 is empty on this site by design: no senolytic has a human lifespan RCT.

Honesty flag

Each intervention carries a flag derived by string classification of its human-evidence field: ● available is reserved for a genuine human lifespan/clinical-outcome RCT; ▲ qualified marks human data that is pilot, early-phase, ongoing, failed, or observational — i.e. not a lifespan RCT; ○ none marks animal/preclinical-only or mechanism nodes. The classifier generalizes across the longevity cluster with zero data fabrication.

Verification status: two levels

Verification is recorded at two levels that are never merged into one field. At record level a study is either verified against its primary publication, verified_as_planned for a registered trial that has not published results, or verified_review_level where the source is review literature rather than a primary report. Separately, each identifier carries its own status: verified, not_independently_verified where the identifier is reported but was not looked up in the registry, not_applicable where it cannot apply to the study type — a trial registry ID for an animal study, for instance — and not_reported where it could apply but has not been confirmed.

There is no “pending” state. It was removed on 2026-08-08: a single field had been carrying both levels at once, so an ongoing-but-registry-verified trial such as AFFIRM was displayed as though its primary source were still outstanding. A missing identifier now says why it is missing rather than promising that someone will come back to it. No identifier is invented at any level.