How EvidenceLib is built

EvidenceLib is a scientific encyclopedia whose articles are assembled from permitted public sources and checked by software before they are published. This page explains what that does and does not guarantee.

Sources and licences

• NLM Medical Subject Headings (MeSH). Definitions, synonyms and classification. U.S. government work in the public domain; the NLM asks for attribution.
• NCBI PubMed (E-utilities). Publication counts and bibliographic identifiers.
• Europe PMC. Abstracts are reproduced only for articles whose licence is Creative Commons Attribution (CC BY or CC0), with the authors credited. Other abstracts and all full texts are linked, never copied.

The pipeline

1. Topics are chosen from MeSH branches for twelve fields and ranked by amount of literature.
2. Facts are retrieved and normalised; duplicates and synonyms of an existing topic are removed.
3. Records are composed from those facts. Nothing is invented: counts, titles and authors come directly from the sources.
4. Every reference is resolved in PubMed; title, year and DOI must match.
5. A quality gate checks definition length, licensing, number of references and total substantive content. Articles that fail are not published.
6. Published articles are rendered to static HTML and listed in the sitemap.

Review states and updates

Articles currently carry the state automated source verification. That means the sources and citations were verified by software; it does not mean a clinician or subject expert has reviewed the text. When an expert review is recorded it is shown on the article, and reviewed articles are locked so that regenerating content cannot overwrite them. Each article has a version number and last-updated date.

Limits

Publication counts show how much has been written, not how strong the evidence is. Open-access abstracts are illustrative samples, not systematic reviews. Nothing here is medical advice.

Corrections

Email support@evidencehelix.com with the article address and the problem.