Validation

AI can organise and explain the evidence. It does not create the evidence.

Depth without discipline is just more numbers. These are the standards that make an integrated report trustworthy rather than merely impressive.

01

Laboratory and assay

Sequencing and biomarker measurement are performed in accredited laboratory settings, with reference materials and external quality assessment used to characterise performance, and confirmatory methods applied where a finding warrants it.

  • Accredited laboratory environments
  • Reference materials for performance characterisation
  • External quality assessment participation
  • Confirmatory testing for clinically weighty findings
02

Interpretation sources

Variant interpretation draws on established public evidence resources rather than internal opinion, and each statement records which source and which version produced it.

  • Clinical variant and population frequency databases
  • Published association and polygenic score catalogues
  • Pharmacogenomic guideline resources
  • Version stamped on every interpretation
03

The longevity signature

Each of the 183 variants is traceable to the replicated literature behind it, carries an effect size and sits in a confidence tier. Markers that do not meet the bar are shown but excluded from scoring.

  • Study-level traceability per variant
  • Effect size published with the finding
  • Four confidence tiers, reviewed on new evidence
  • No over-claiming from small effects
04

Software and AI

Language models organise, connect and explain evidence. They do not generate values, invent findings or decide clinical significance. Outputs are grounded in the underlying measurements and the cited sources.

  • Grounded generation — no invented values
  • Deterministic calculation of all reported figures
  • Model and prompt versions recorded per report
  • Reporting practices aligned to recognised AI reporting frameworks
05

Medical oversight

Where a finding requires medical interpretation, it is reviewed by a qualified professional before release, and escalation paths exist for results that should not wait.

  • Professional review where interpretation is medical
  • Defined escalation for critical values
  • Clinical decision support augments a clinician, never replaces one
  • Referral to appropriate care rather than treatment advice

These describe the standards Aevum builds to. Aevum does not claim certifications, regulatory approvals, contracted clinical partners or advisory appointments that have not been formally confirmed.

Limits

What an Aevum report cannot do

Stating this plainly is part of the product, not a footnote to it.

It does not predict how long you will live.
It does not diagnose disease.
It does not prevent disease or reverse ageing.
It does not replace your doctor.
It does not guarantee any health outcome.
It does not turn small genetic effects into certainties.