PubMed / PubMed Central (PMC)

PubMed is the reference index for biomedical and life sciences literature, run by the US National Library of Medicine, part of the National Institutes of Health. It holds more than 36 million citations spanning medicine, nursing, pharmacology, and related fields. Each record gives you a citation and, where available, an abstract, along with structured Medical Subject Headings (MeSH) tags: a controlled vocabulary you can use to filter a corpus down to a precise topic.

PubMed Central (PMC) is the companion archive that stores free full-text articles rather than just citations and abstracts. The part you want for RAG is the PMC Open Access Subset: several million articles you are allowed to download, process, and redistribute at scale. Both the index and the full text are available in bulk over FTP as XML, and the Entrez E-utilities API is there for smaller, live lookups by term or identifier.

This is the foundation for any biomedical or clinical RAG system. The MeSH tags make topic filtering straightforward, abstracts alone are often enough to ground a factual answer, and the Open Access full text lets you retrieve at the passage level when you need depth. Building on the Open Access subset rather than the whole of PMC keeps you on safe legal ground.

The licensing needs care. Citation metadata is public domain, but full-text article licences vary article by article: much of the Open Access subset is Creative Commons, yet some entries carry stricter non-commercial or no-derivatives terms, so check the licence on each article before commercial reuse or redistribution. That is why this entry marks the content licence as differing from the metadata.

Alongside arXiv, where biomedical preprints increasingly appear too, and Semantic Scholar's broader citation graph, PubMed is the specialist backbone for anything health-related.