ClinicalTrials.gov
ClinicalTrials.gov is the official US registry of clinical research studies, run by the National Library of Medicine at the National Institutes of Health. It holds structured records on more than 450,000 studies from around the world. Each record follows a consistent structure: the condition being studied, the intervention being tested, eligibility criteria, study design, locations, sponsors, and, where available, reported results and adverse events.
The practical way in is the version 2 REST API, which returns clean `JSON` and lets you page through results by condition, sponsor, status, or free-text query. For a whole-corpus build there is a bulk download of the full dataset, plus per-study `XML` and `CSV` exports. A common pattern for RAG is to keep the structured fields as metadata (phase, status, condition, dates) and turn the free-text sections, the summary, eligibility, and outcome descriptions, into the chunks you embed, so filters and semantic search work together.
It is a strong fit for RAG that answers questions about treatments, ongoing or completed research, trial design, or whether a patient profile might match a study's eligibility. The consistent schema means you can build reliable filters, and being public domain (a US Government work) you can reuse everything freely, with no attribution or share-alike obligations.
Two caveats worth planning for. Records are self-reported by sponsors and vary a lot in completeness: many trials never post results, so absence of an outcome is common rather than meaningful. And while registration is required by law in many cases, which keeps coverage broad and current, registration is not proof of quality or of a published finding.
Pair it with PubMed or Semantic Scholar for the published literature behind a trial, and with DrugBank or ChEMBL when you need detail on the drugs being tested.
Related sources
AlphaFold Protein Structure Database
EMBL-EBI and Google DeepMind's open database of AI-predicted protein structures, covering over 200 million sequences across almost all of UniProt. Each model ships with per-residue confidence and predicted aligned error scores, and is reachable by website, API, FTP and Google Cloud bulk access. Released under CC BY 4.0 and designed to pair with UniProt and the experimental structures in the PDB.
DrugBank
A freely accessible database that combines detailed drug data with drug target information, covering more than 15,000 drug entries. It links chemistry, pharmacology, and biology in one place, which makes it a strong knowledge source for medical and pharmaceutical RAG.
MIMIC-III / MIMIC-IV
Freely accessible critical care databases with de-identified health data from more than 40,000 intensive care patients at Beth Israel Deaconess Medical Center. They include vital signs, lab results, medications, and clinical notes. Access needs credentialing and a data-use agreement.
MIRIAD
A million-scale medical instruction and retrieval dataset of roughly 5.8 million question-answer pairs, each grounded in a passage from peer-reviewed biomedical literature. Built for medical RAG, retrieval, and hallucination detection.