Data Sources, Licensing and Limits of the Library
Where Research Library records come from, what imported data can and cannot tell you, how abstracts and licenses limit reuse, and what the library does not contain.
Content last updated . This page has not been medically reviewed.
The short answer
The Research Library is built from records imported from public databases and ontologies. It is not a set of articles written and checked one by one. That is useful for finding what exists, and it comes with limits: imported records can be incomplete, out of date or wrong, and none has been reviewed by a clinician.This page explains what the library holds, what stays outside it, and how licensing shapes what appears. For how to read an individual record, see how to read Research Library records.
What the library holds
| Record type | Approximate count | Notes |
|---|---|---|
| Herbs | 6,800 | Names, identifiers, linked study records |
| Ingredients | 44,600 | Same |
| Formulas | 6,700 | Same |
| Condition and syndrome terms | 30,500 | Terms from ontologies, not diagnoses |
| Symptom (phenotype) terms | 19,900 | Same |
| Gene records | 15,700 | Official symbols and identifiers |
| Biomarker records | 25 | Measured characteristics |
| Registered clinical trials | 8,558 | 6,881 have status Completed |
| Systematic-review registrations | 8,032 | 833 recorded as completed; 6,877 recorded as ongoing |
| Publication records | 6,705 | With PubMed IDs and abstracts |
Most records have names, identifiers and linked study records but no written summary. Research model labels on publication records (animal, cell, clinical) come from the source data.
Types of identifiers you will see
Study records carry identifiers that let you find the original. Search supports exact lookup by DOI, PubMed ID, NCT number and CRD number in the evidence collection, so you can search by identifier.A DOI is a unique number made up of a prefix and a suffix separated by a forward slash, and it is a persistent, long-lasting reference to an object. [1] A PubMed ID identifies a citation in PubMed, the free NLM resource that holds citations and abstracts of biomedical literature. [2] Search also recognizes NCT numbers and CRD numbers, which belong to the trial and systematic-review registrations.
What a publication record is and is not
PubMed contains citations and abstracts and does not include full-text journal articles, although links to the full text are often present. [2] A publication record in the library therefore stands for an abstract-level description of a paper. An abstract is only a short summary of a paper. Open the original paper before relying on any detail, and see evidence types and study designs.Imported data can lag or contain errors
Data providers revise their records. NLM, for example, produces daily files that include new, revised and deleted PubMed citations, and says that revised or deleted citations should replace existing ones in a local copy. [3] A copy of any database taken at one time may therefore not show later changes. This includes corrections, expressions of concern and retractions. PubMed links errata and expressions of concern through fields such as "Erratum in" and "Expression of concern in." [4] The page on retractions, corrections and duplicate reports explains how to check for them.Data providers also limit what they promise. NLM's terms for its downloadable data say NLM makes no warranties, express or implied, about the accuracy or completeness of the data. [5] Within the library, connections found by name matching can be wrong, and a link imported from another database has not been checked here.
Licensing and reuse
Some data are free to reuse and some carry conditions. NLM asks users of its downloaded data to acknowledge NLM as the source with the words "Courtesy of the U.S. National Library of Medicine." [5] Publishers may also attach a copyright statement to a record. [4] Because of this, the library points you to the original rather than reproducing copyrighted full text. When you quote or reuse material, check the license of the original source, and see citing and exporting library records.What is missing
- Written summaries. Most records have none.
- Clinical review. None of the imported records has been reviewed by a clinician.
- Full study text. Publication records carry PubMed IDs and abstracts, not full papers.
- Ratings. The library shows connections and does not rate effectiveness.
- Complete coverage. The records reflect what the source databases hold. A herb with few records is not necessarily less studied in the world, only less represented here.
- Phase for every trial. A phase is recorded for about two thirds of trials; the rest have no phase.
Search engines and thin records
Records are indexed by search engines only when they have linked study records or a consumer guide. Thin imported records stay on the site but are kept out of search-engine results. This is about visibility, not quality.Frequently asked questions
Who wrote the records?
They were imported from public databases and ontologies. They are not editorially written profiles.
Has a clinician checked the data?
No. None of the imported records has been reviewed by a clinician. The default verification state is "imported."
Does the library include full-text studies?
No. Publication records carry PubMed IDs and abstracts. PubMed itself does not include full text articles. [2]
Why might a record be out of date?
Source databases revise and delete records over time, and a copy may not reflect later changes. [3]
References
- International DOI Foundation: What is a DOI?. Read 2026-10-22.
- National Library of Medicine: About PubMed. Read 2026-10-22.
- National Library of Medicine: Download PubMed Data. Read 2026-10-22.
- National Library of Medicine: PubMed User Guide (Help). Read 2026-10-22.
- National Library of Medicine: Terms and Conditions (downloaded data). Read 2026-10-22.