Define the object
Every indexable evidence page represents an Intervention or Study. We do not create a URL for a vague keyword theme.
Evidence method · v0.1
How Leptox reads evidence.
The data model keeps the intervention, study, population, endpoint, effect, evidence level, and commercial status distinct.
Extract the chain
Results are stored with their population, comparator, endpoint, and time horizon.
Grade the claim
Evidence level reflects directness, trial design, consistency, scale, and applicability, not popularity.
Preserve nulls
Non-significant and conflicting findings remain visible. They are not edited away to create a cleaner story.
Separate commerce
Commercial availability describes access. It cannot alter the evidence grade or interpretation.
Review the record
Each release carries one evidence-review date. Material corrections update the source data and regenerate every affected surface.
Evidence levels
High
Large randomized trials directly measure the primary outcome in a relevant population, with a consistent direction of effect.
Moderate
Randomized evidence supports an effect, but magnitude, generalizability, durability, or intervention definition remains context-dependent.
Low to moderate
Multiple trials or syntheses exist, but results are small, inconsistent, heterogeneous, or vulnerable to product and study-quality differences.
Scope limits
Leptox is an educational evidence reference. It does not diagnose, prescribe, recommend doses, or replace individualized care. Regulatory status, product labeling, and evidence can change; source links are provided for verification.
Corrections: factual corrections should be applied to the authoritative data record, then propagated through the generator and validation pipeline.