PhysioNet
PhysioNet is maintained by MIT's Laboratory for Computational Physiology and has spent more than two decades as a home for reproducible medical research data. Its catalogue runs from physiological waveforms such as heart traces (ECG) and brain activity (EEG) to structured clinical databases, medical imaging, and the open-source software used to analyse them. It is also where several landmark datasets are distributed, most famously the MIMIC critical- care databases of de-identified intensive-care records.
You work through the catalogue project by project: browse by topic, open a project page, and download its files, documentation, and licence from one place. Formats vary a lot between projects, from CSV tables and free-text clinical notes to specialised waveform files, so expect to write a loader per dataset rather than a single pipeline. For a clinical RAG system, the text- rich datasets are the obvious raw material, the discharge summaries, radiology reports, and progress notes inside MIMIC among them. Clean and de-identify the notes, chunk them sensibly, and embed them next to the structured fields you want to filter on.
PhysioNet earns its place when you need genuine clinical language and physiological measurements rather than general medical prose. It suits research prototypes, clinical-decision-support experiments, and any system that must reason over real, de-identified patient records.
Access is the thing to plan for. Plenty of datasets download openly, but the sensitive ones, MIMIC included, require you to become a credentialed PhysioNet user, which means completing recognised human-subjects research training and signing a data-use agreement that can take days or weeks to clear. Licences differ from one project to the next and several are research-only, so commercial use is not a given: read the terms on each page before you build, and never pass the underlying records to a third-party service in a way the agreement forbids.
For openly licensed drug and trial metadata to sit alongside it, ClinicalTrials.gov and DrugBank complement it well. PhysioNet is the source to reach for when you specifically need patient-level clinical records and signals.
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.
ClinicalTrials.gov
A US National Institutes of Health database of privately and publicly funded clinical studies. It holds structured records on more than 450,000 studies, including protocols, conditions, interventions, outcomes, and results, all in the public domain.
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.