Data Standards
How we build the compound library
Every compound profile, citation, and study reference on Staqk is sourced directly from peer-reviewed literature indexed in the National Library of Medicine. No editorialising. No vendor partnerships. Just the published science.
One source: NLM PubMed
PubMed is the National Library of Medicine's database of over 36 million biomedical citations and abstracts. It is maintained by the US National Institutes of Health and is the reference standard for clinical and life-sciences literature globally.
We use only PubMed as our reference source. This means every paper linked in a compound profile has been submitted to, reviewed by, and indexed within an independent government-maintained scientific database — not curated by us.
How a paper enters the library
PubMed search
We query the NLM E-utilities API with each compound name and any known aliases. Only English-language publications are retrieved.
Metadata retrieval
For every result, we pull the exact paper title, full author list, journal name, publication year, and DOI directly from NLM's esummary endpoint.
Abstract extraction
Where available, we fetch the full abstract via efetch. Structured abstracts — those with labeled sections — are joined into a single readable block.
Publication type classification
NLM tags every paper with one or more publication types. We map these to six study tiers: RCT, clinical trial, meta-analysis, observational, case report, and review.
Automated title verification
A separate verification pass compares stored titles against live NLM data using a hybrid Jaccard + Jaro-Winkler similarity score, flagging any drift.
Linked to compound profile
Each paper is linked directly to the relevant compound. Every reference includes a live PubMed URL so you can read the source in full.
Study type classification
NLM assigns publication types to every paper it indexes. We map those tags directly to one of six study tiers — no manual classification.
Participants randomly assigned to intervention or control. Highest standard for establishing causality.
Prospective study in human subjects. Phase I–IV trials, dose-finding studies, and safety assessments.
Pooled analysis across multiple studies. Synthesizes evidence and quantifies overall effect sizes.
Cohort, case-control, and cross-sectional designs. No intervention — documents outcomes in real-world populations.
Detailed account of an individual patient. Useful for documenting rare effects but cannot establish prevalence.
Narrative summary of existing literature. Valuable for orientation but not a primary data source.
What we don't include
- Vendor-produced content or sponsored research
- Papers not indexed in NLM PubMed
- Retracted publications
- Anecdotal community reports presented as clinical evidence
- Content that promotes a specific product or supplier
Important limitations
This is not medical advice. The compound library is an educational reference. Nothing here constitutes a treatment recommendation, diagnosis, or clinical guidance. Consult a qualified healthcare professional before starting any protocol.
Research quality varies by compound. Many of the compounds in this library have limited or no human clinical trial data. Animal and in-vitro studies are included where they are the best available evidence, but they are clearly labelled as such.
The science is evolving. PubMed is updated continuously. Papers get retracted, new trials publish, and consensus shifts. We run periodic re-verification passes to keep the library current, but there may be lag between new publications and their appearance here.
Reporting an error
If you spot an incorrect title, a wrong PubMed link, a misclassified study type, or a compound profile that references an unrelated paper, please let us know. We verify every report against the original NLM source before making any change.
hello@staqk.app