Open evidence data
AgeBeacon Open Human Longevity Evidence Dataset
Dataset v2.0 expands AgeBeacon beyond the microbiome into a controlled multi-approach release. It preserves the v1.1 microbiome snapshot and adds only trials with QC-reviewed, manually verified outcome evidence from deeper intervention work. Earlier releases remain preserved as historical snapshots.
87
trial/intervention records
17
direct relevance
70
indirect relevance
Released 2 October 2026 · v2.0 methodology: multi-approach-pipeline-v3 · 65 structured verified outcome records
What changed in v2.0?
v2.0 preserves the microbiome v1.1 evidence and adds QC-reviewed outcome evidence from additional deep/scaled interventions including plasma exchange, rapamycin, NR, metformin, senolytics, Urolithin A, NMN, Acarbose and Resveratrol. Inclusion still does not imply efficacy.
How to read it
Direct means the trial has a clear healthy-aging, frailty, age-related function or closely aligned aging endpoint. Indirect means the trial informs an age-related population or outcome but should not be read as direct longevity evidence.
Inclusion is not evidence of benefit. A registered trial shows research activity. Registry results, publications and efficacy conclusions require separate verification. v2.0 distinguishes registry-posted results from manually verified results publications and includes outcome-level records where extraction has passed verification. Absence of an extracted outcome still means unresolved—not neutral.
Study-quality bands describe design features such as randomization and masking. They do not say whether the intervention produced a beneficial result.
Fields
The download includes intervention, registry source and ID, relevance, title, recruitment status, enrollment, randomization, masking, registry-results availability, verified-results-publication status, study-quality band/version and source URL.
Versioning and corrections
Dataset releases are snapshots from a controlled public release layer. The working ingestion database remains private. Material corrections belong in later versions rather than silently rewriting earlier releases. v1.0 and v1.1 remain preserved as historical releases; v2.0 is the first controlled multi-approach release.