MIMIC-III / MIMIC-IV
MIMIC (Medical Information Mart for Intensive Care) is one of the most widely used open clinical datasets. It collects de-identified records from tens of thousands of intensive care unit stays, covering vital signs, laboratory results, medications, procedures, and free-text clinical notes written by hospital staff. The free-text notes in particular make it valuable for building and testing clinical RAG systems.
There are two generations to choose from. MIMIC-III is the long-established version that underpins a huge body of published research. MIMIC-IV, hosted at https://physionet.org/content/mimiciv/, is the newer release with an updated structure and more recent admissions. Both are distributed through PhysioNet.
Access is restricted rather than fully open. You must become a credentialed PhysioNet user, complete recognised human-subjects research training, and sign a data-use agreement before you can download anything. Build that approval time into your project plan.
Related sources
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.
PhysioNet
A repository of freely available medical research data, including physiological signals such as ECG and EEG, clinical databases, and related software. It hosts MIMIC and many other biomedical datasets, so it is a central hub for clinical RAG source material.
UniProt
A resource for protein sequence and functional information, with more than 250 million protein sequences and rich annotations. It is a core reference for bioinformatics, so it grounds life sciences RAG systems in reliable protein facts.