How Every Dose Is Cited

Most peptide dosing information online comes from forums, vendor marketing, or copied blog posts — with no traceable source. We built PeptideDosageCharts to be the opposite: every dosing figure is anchored to a specific, verifiable primary study.

1. Literature retrieval

For each compound we query PubMed via the NIH/NCBI E-utilities API, targeting dosing, administration, and pharmacokinetic literature. We retrieve the most relevant studies and their full abstracts.

2. Structured extraction

A language model reads only the retrieved abstracts and extracts dosing protocols (goal, dose range, frequency, duration, route). It is explicitly constrained to extract only figures stated in or directly supported by the source text — and to map each protocol to the PubMed ID(s) that support it.

3. Anti-fabrication filter

Any protocol that cannot be anchored to a retrieved, real PubMed citation is discarded, not published. A blank cell is acceptable; an invented number is not. Each citation URL is verified to resolve before it is stored. This is why some compounds show fewer rows than you might find elsewhere — we will not publish a dose we cannot source.

4. Evidence labeling

Every row carries an evidence level — clinical, preclinical, anecdotal, or consensus — so you can weigh human-trial data differently from animal or in-vitro findings.