How claims are built, labelled, and retired
The value of a mechanistic framework depends entirely on whether its labels are honest. This page states the rules the rest of the site is held to.
Last materially reviewed
Source hierarchy
Primary peer-reviewed literature is preferred over reviews, reviews over commentary, and commentary over secondary reporting. Where a claim rests on a specific paper, that paper is cited by PMID or DOI in the claim registry rather than described second-hand.
Mechanistic work in animal or cell systems is cited as such and never re-labelled as human evidence. A source is not upgraded because the conclusion is convenient.
Mechanistic inference versus direct evidence
Most of what makes BiomeLogic distinctive is integration: connecting established component biology into a reading of a clinical pattern. That integration is inference, and it is labelled as inference everywhere it appears.
A claim built from established parts does not inherit their evidence tier. The parts stay established; the integration stays inference until something tests it directly.
Association versus causation
Directional language is reserved for steps where a causal relationship is demonstrated in the cited work. Where the literature shows co-occurrence, the site says association and the evidence map marks the relationship as associative rather than causal.
No observational material — including any future case synthesis — is used to argue that an intervention caused an outcome.
Handling contradictory evidence
Contradictions are published, not resolved by omission. Where credible work conflicts with the framework, it is recorded as an unresolved contradiction and surfaced alongside the claim it challenges.
Competing explanations for the same observations are listed on the framework page itself so a reader can see the alternatives without leaving the argument.
How uncertainty is labelled
Every claim carries one tier from a fixed vocabulary. The tiers are ordered, defined once, and reused across the claim registry, the evidence maps, and the public API — so a label means the same thing everywhere on the site.
- strong-human-evidence
- Strong human evidence
- animal-mechanistic-evidence
- Animal / mechanistic evidence
- in-vitro-evidence
- In vitro evidence
- ecological-microbiome-evidence
- Ecological / microbiome evidence
- clinical-pattern-inference
- Clinical-pattern inference
- hypothesis
- Hypothesis
- speculative
- Speculative
How hypotheses are generated
A hypothesis is written only when it is falsifiable: it must specify what observation would weaken it. Hypotheses appear under an explicit heading and are never presented as findings, guidance, or expected results.
How case observations differ from clinical evidence
A single situation, however detailed, is a description. It cannot establish efficacy, cause, or generalisability. Case material is therefore held to a separate template with an explicit list of conclusions it does not support, and it is never used to strengthen a mechanistic claim.
Corrections and updates
Material revisions carry a visible review date. When a claim is weakened or withdrawn, the change is recorded rather than silently edited away, and pages that depended on it are revisited.
Scope limits
BiomeLogic produces independent systems-biology analysis and writing. It does not diagnose, prescribe, treat, or replace medical care, and it does not interpret an individual's results as a clinical decision. Anything actionable belongs with a licensed clinician.