Evidence Grading
UKS treats evidence quality as first-class and never collapses it to a boolean. Two independent, closed vocabularies travel with every source: an evidence grade and a clinical status.
Evidence grade (GRADE-aligned)
| Grade | Meaning |
|---|---|
high | Meta-analyses of RCTs |
moderate | Individual RCTs with limitations |
low | Observational / cohort studies |
very_low | Expert opinion |
n_a | Datasets, non-clinical content |
This mirrors the GRADE framework used in evidence-based medicine. It is a closed vocabulary — mis-grading evidence is a safety hazard, so new values require a minor version bump, not an ad-hoc extension.
Clinical status
| Status | Meaning |
|---|---|
validated | Established consensus / standard of care |
emerging | Promising, not yet replicated at scale |
contested | Actively disputed |
commercial | Primarily marketing content |
informational | Factual, non-clinical |
n_a | Not applicable |
Clinical status is separate from evidence grade — a source can be high-grade and contested at the same time.
Why "contested" matters
Some claims are genuinely disputed, and flattening them to "true/false" is dangerous. For example, genotype-guided dosing for MTHFR / COMT / APOE must be marked contested, per ACMG guidance. The Registry enforces this at publish time: asserting such content as validated is rejected. The registry can also derive contestedness empirically from the corpus — see Trust & safety.
Licenses are never collapsed either
Every source carries both a license_label (e.g. "CC BY 4.0") and a rights_url. This pairing lets an agent make a correct reuse decision instead of guessing from a single boolean "open?" flag.