by PhysioNet
PhysioNet is an NIH-funded platform run by the MIT Laboratory for Computational Physiology that hosts curated, de-identified biomedical and clinical datasets, software and models. Most resources are openly available; some require credentialing.
DatasetFree
Step 1: Search or browse published projects on physionet.org and open a project page.
Step 2: Check the Files section for the license and any access requirements.
Step 3: For credentialed data, create an account, complete credentialing and the required training, then sign the data use agreement.
Step 4: Download the files, or install the physionet package and run physionet download with the project name and version.
Step 5: Cite the dataset authors and the PhysioNet papers listed on the About page when you publish.
Credentialing is a review of your identity and research interest by the PhysioNet team. Training is proof you completed a course required by the dataset. Not all datasets require both, so check the project's Files section.
Approval typically takes 1-2 weeks but can vary. Applications are reviewed in the order received, and an institutional email or linked ORCID profile helps avoid delays.
Use the physionet Python package, which can download a whole project or selected files, resume interrupted downloads and verify checksums. Open datasets may come from an AWS mirror, which is recommended for faster downloads.
Yes. You create a project, submit it for editorial review and, once accepted and approved by all authors, it is published. You choose a license and can run physionet validate before submitting.
KaggleKaggle is a Google-owned data science platform where users find and publish datasets, run code in a browser-based environment and enter machine learning competitions.CT-RATECT-RATE is a dataset of chest CT volumes paired with radiology text reports, multi-abnormality labels and metadata. It is for researchers working on 3D medical imaging and vision-language models.