Sharing on PhysioNet

Why share on PhysioNet

  • Visibility. Published projects are indexed, versioned, and citable, which makes your work easier to find and credit.
  • Collaboration. Others can build on your work and contact you about it.
  • Reuse. Clear documentation and standard formats help others reproduce and extend your results.
  • Transparency. Editorial review encourages higher-quality, better-documented research outputs.

What PhysioNet publishes, and what it does not

PhysioNet publishes resources: the data, software, or models that others can reuse. We typically do not review and publish the results of scientific findings. Report those in a journal or conference paper and cite that paper from your project.

Before submitting your work, please review our author guidelines.

Submission Overview

Choosing a project type

When submitting a project to PhysioNet you will be asked to select one of the following project types:

  • Database: Research data with significant potential for reuse by the research community. This may include data that enables published studies to be reproduced, data for benchmarking algorithms, and data that supports novel investigations.
  • Software: Software that has been developed for research applications.
  • Model: An implementation of a statistical or machine learning model with potential for reuse by the research community. Typically models are created by a training process and may have dependencies on specific computational frameworks.
  • Challenge: A description of a challenge for the research community. Files such as datasets and software may be included as part of the challenge.

How the process works

  1. If you are the submitting author, create a new project using the project management system.
  2. Add your project details to each tab of the submission: Authors, Content, Access, Discovery, Ethics, Upload Agreement, Files.
  3. Review everything in Proofread, then submit.
  4. The editorial team reviews your project and sends an initial decision within six weeks.
  5. If accepted, your project will be copyedited and sent to the authors for review and final approval.
  6. Your project is published. Published content is fixed and cannot be changed. To make changes, publish a new version.

Before you upload, we recommend these two tools:

Before submitting

Before submitting your project for review, please make sure that you have:

  • Carefully read the author guidelines for each section of the project.
  • Read the PhysioNet Deposit Terms.
  • Selected an appropriate license for releasing your data files.
  • Attached your IRB or ethics committee approval (or exemption determination) if you collected new human or animal data.

In addition, please ensure:

  • Your project content and files are clearly described.
  • Your files do not contain any protected health information.
  • Details of the data collection process are provided in the Methods section. Supporting information and diagrams should be provided for less well-known data capturing devices.
  • Relevant publications are included in the References section and are cited in the main text.
  • If this is an updated version of a previously published resource, the major changes from the previous version are clearly specified in the Release Notes.

Author Guidelines

The sections below give detailed instructions for each tab of the submission portal. Each one lists what is required and what is recommended.

  • Required items must be included whenever they apply to your project. For example, software projects do not need to report a cohort size. Projects missing required items will be sent back for revisions or rejected.
  • Recommended items are what we suggest for a high-quality project. They won't normally hold up publication, but editors may ask for one if your project can't be fully understood or reused without it.

Our review aims to ensure that each resource is reusable, free of protected health information, and described well enough that others can understand it and use it correctly.

1. Authors

Inviting co-authors. In this section the submitting author must invite each co-author to the project. Co-authors will receive a notification message upon submission and must approve the submission prior to publication.

Setting affiliations. Each author must set their affiliation for the project (Institution & Country). Each author may set up to three institutions for the project. Authors must provide their affiliation on a project-by-project basis.

Authorship vs acknowledgment. Authors are people who made substantial contributions to creating the resource and who accept responsibility for it. People who helped but don't meet that bar belong in the Acknowledgments field.

2. Content

This section holds the written description of your resource. Treat it as a short data paper: it should let a reader understand the context and reusability of your resource, decide whether the resource suits their needs, understand how it was made, and use it correctly without contacting you.

Rules that apply to all Content fields

  • The description must be self contained. You may link to a project website in the Discovery tab, but the documentation on PhysioNet must be sufficient to understand and use your resource.
  • No URLs or hyperlinks in the main text. Add every external resource (websites, publications, datasets, GitHub repositories) to References and cite it by its key.

Title

Required

  • Is no more than 200 characters.
  • Contains only letters, numbers, spaces, underscores, and hyphens.
  • Does not start with “The”.
  • Avoids acronyms and abbreviations where possible.
  • If derived from MIMIC and you want to use the MIMIC name, include “Ext”, as in MIMIC-IV-Ext-YOUR-DATASET. “Ext” can mean extracted (a derived subset) or extended (for example, added annotations).
  • If the dataset is derived from another dataset, the title must make this clear.

Recommended

  • Say what the resource is: the data type, population, and purpose.

Abstract

A brief description of the resource and the context in which it was created. It may appear in search indexes independently of the full project metadata, so providing detailed information about the content is important.

Required

  • No more than 250 words.
  • No references.
  • Describes the resource itself, at a high level, and the key aims of the project.

Recommended

  • If the resource was generated as part of a scientific investigation, include relevant information to facilitate reuse.
  • Give concrete numbers: number of participants or records, signal types, time span, setting, and the access level.

Background

Required

  • Introduce the resource and the context in which it was created.
  • Explain your motivation for sharing it.
  • Say what gap the resource fills and how it relates to existing resources.
  • Provide references for any claims or comparisons.

Recommended

  • A short comparison with related resources that distinguishes why your resource is unique.
  • A description of parent resources and what this resource adds.

Background explains why the resource exists. Usage Notes explain how to use it.

Methods & Technical Implementation

Required

Provide all information needed to understand, reproduce, and use the resource, including the procedures used to create it. Include the items below that apply to your project type.

  • Data: where applicable, describe the experimental design, cohort size, measurement devices, any processing or transformations applied, inclusion and exclusion criteria, how annotations were produced (number of annotators, guidelines, disagreement resolution), de-identification techniques, and any use of LLMs (model and version).
  • Software: development process, software design, and description of algorithms.
  • Models: the training procedure, the training data, the framework, and hyperparameters.

Recommended

  • Derived data: the exact source dataset and version, and the transformations applied.
  • An example prompt for LLM usage.

Content Description

Required

  • How the files are organized, the file formats, and what each file or group of files contains.
  • Tabular data: a data dictionary giving, for each variable, its name, type, unit, allowed values, and how missing values are coded. This may also be included as a separate CSV file.
  • Software and models: the package layout, entry points, inputs, outputs, and required hardware.

Recommended

  • Summary statistics where appropriate: number of patients or participants, number of files, signal types, and time span.
  • Relationships between files, such as the keys that link tables.
  • Waveforms: signal names, units, sampling frequency, and the number of channels per record.
  • An annotated directory tree.

Usage Notes

Required

  • The reuse potential of the resource.
  • Known limitations users should be aware of.

Recommended

  • How the data has already been used (citing relevant papers).
  • Complementary code or datasets that might be of interest to the user community.
  • Example code or a notebook that loads the data.
  • Suggested train, validation, and test splits.

Release Notes

Required

  • For new versions of projects published on PhysioNet: Specify major changes from the previous version.

Recommended

  • For new projects: A statement that the current version is the initial release of the project (e.g., “Version 1.0.0 – Initial Release”).

Acknowledgments

Recommended

  • Funding information.
  • People who contributed but are not authors.

Conflicts of Interest

Required

  • A statement on potential conflicts of interest.
  • If the authors have no conflicts of interest, write: “The author(s) have no conflicts of interest to declare.”

References

Required

  • A list of citations used in the project in APA style.
  • Cite each reference by its key, for example [@moody2001], or [@moody2001; @mark1997] for several. Each citation links to its reference on the project page.
  • Websites, datasets, and code repositories belong in the reference list as well and should be cited properly in APA style and not just as hyperlinks.

Recommended

  • If a reference has a DOI, enter it and click the Look up button to automatically fill in the citation.
  • List references in the order they are first cited. This order is shown in the preview and applied automatically when the project is published

3. Access

This tab sets who can access the files and under what terms. When sharing data and software, it is important to be clear about how you intend the content to be reused. To maximize the reuse potential of the content, we encourage permissive, open licenses.

Choosing an access level

Access level Who can download Typical use
Open Anyone Data with minimal privacy risk, and software
Restricted Registered users who sign a Data Use Agreement (DUA) More sensitive data that needs a DUA but not individual vetting
Credentialed Users who have been credentialed and have signed the DUA Sensitive de-identified clinical data
Contributor Review PhysioNet credentialed users who sign a DUA and submit a project plan approved by the data contributor Data that is only available for specified research uses

PhysioNet editors reserve the right to require a more restrictive access level if they deem the data too sensitive for open sharing.

Licenses by project type

License Access Policy Database Software Challenge Model
Open Data Commons Attribution License v1.0 Open ✓ ✓
Creative Commons Attribution 4.0 International Open ✓ ✓
Creative Commons Attribution-ShareAlike 4.0 International Open ✓ ✓
Open Data Commons Open Database License v1.0 Open ✓ ✓
Creative Commons Zero 1.0 Universal Open ✓ ✓
Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Open ✓ ✓
GNU General Public License v3 Open ✓ ✓ ✓
MIT License Open ✓ ✓ ✓
3-Clause BSD License Open ✓ ✓ ✓
PhysioNet Restricted Health Data License 1.5.0 Restricted ✓ ✓ ✓
PhysioNet Credentialed Health Data License 1.5.0 Credentialed ✓ ✓ ✓
PhysioNet Contributor Review Health Data License 1.5.0 Contributor Review ✓ ✓ ✓

Derived projects

If your project is derived from another dataset or model:

  • Your license and access level cannot be less restrictive than the source allows. For credentialed sources such as MIMIC, derived resources are shared under the same agreement as the source data.
  • Your title must make the relationship clear (see Title).
  • Select the parent project in the Discovery tab.

4. Discovery

Discovery fields help people find, cite, and place your project in context.

Field Required Notes
Version Yes Use semantic versioning (major.minor.patch). If unsure, enter 1.0.0.
Short Description Yes This should be one or two complete sentences that describe the contents of the project to a reader who is generally knowledgeable about the subject but not specifically familiar with your research.
Project Home Page No External home page for the project
Parent Projects If derived Select the existing PhysioNet project(s) this resource was derived from.
Publication For most ML models Journal publication that this resource is associated with.
Topics Recommended Keyword topics associated with the project. Increases the visibility of your project.
Funding Recommended Each organization that funded this work, with the award or grant number if there is one. Add a separate entry for each award.

DOI. A DOI is assigned when the project is accepted. If you need the final citation, including DOI, for an accompanying paper, contact the editorial team after acceptance and before publication.

Citations. Published project pages show a “Cited by” list of publications that cite the project. To be counted, cite the project in work that uses it.

Sharing with article reviewers. Once your project is submitted, you can request a temporary private link to share with reviewers of a related article from your project submission page.

5. Ethics

The ethics statement should cover benefits, risks, and the approvals behind your work.

Required for all projects

A statement on the ethics of your work. Briefly describe the benefits and the risks of sharing it.

Required for projects that collect new data from human subjects or animals

Attach an official approval document confirming that the study protocol was reviewed and approved by the submitting institution's IRB or ethics committee. If your study was determined to be exempt from review, attach the documentation of that exemption determination instead. Attached documents are visible only to the PhysioNet editorial team.

This requirement does not apply to projects that are derived entirely from existing datasets, including those that add annotations or other derived data to them. These projects should instead describe the source data's ethical approval in the ethics statement (see the example wording below).

Required for the following, where they apply

If your project involves… State
Human subjects That the study protocol was approved by the relevant Institutional Review Board (IRB) or ethics committee, with the name of the board and the approval or protocol number where possible. State whether subjects gave written informed consent, or whether consent was waived.
Clinical trial data The trial registration number and the name of the registry.
Animal data The animal care guidelines followed and the approving body. See, for example, this project and this official NIH manual.
Data from another source That you had permission to redistribute it, and under what terms.

Recommended

  • Confirm that consent or approval covers public or controlled sharing of the data, and not only the original collection.
  • Describe the steps taken to prevent re-identification and any residual risk.

Example Wording

Human subjects: “This study was approved by the [Board name] (protocol [number]). Written informed consent was obtained from all participants, including consent to share de-identified data publicly. All dates were shifted and direct identifiers removed (see Methods).”

Derived from MIMIC: “All data in this project are sourced from MIMIC-IV and no new data are introduced. MIMIC-IV was reviewed by the Institutional Review Board at Beth Israel Deaconess Medical Center, which approved the data-sharing initiative and granted a waiver of informed consent.”

6. Upload Agreement

Before submitting, you must agree to the PhysioNet Upload Agreement and the Deposit Terms. In summary, by agreeing to the Upload Agreement, you confirm that you have the right to share the content and the files contain no personally identifiable information. You must select whether your project contains any data from human subjects or whether it contains data derived from other de-identified datasets published on PhysioNet or elsewhere.

7. Files

PhysioNet’s goals are to ensure that the content is safe to share and that it is sufficiently well structured and described for it to be a valuable resource for the research community. When submitting a project, you will be asked to upload relevant data and software files.

Rules for all projects

README file

  • Required: A README file at the top level, with a title and a brief description of the package contents
  • Recommended: The directory layout, how to cite the project, the license, the version, and a contact.

Protected health information (PHI)

  • Required: All PHI must be removed. All dates (except year), including data collection dates, must be date-shifted or removed. Ages over 89 must be aggregated into a single value (e.g., “90+”). Please refer to the comprehensive guide for de-identifying data for more information.
  • Recommended: Check free-text fields and clinical notes, device and record identifiers, and file metadata.

File naming

  • Required: Name files clearly. No spaces (use underscores) and no special characters.
  • Recommended: Use lowercase filenames (exceptions are “special files”, such as the RECORDS and ANNOTATORS files used for waveforms). Use shorter filenames (e.g., 1001abp.dat is better than subject_1001_ABP_wave_100Hz.dat).

File formats

  • Required: Use open, machine-readable formats. Proprietary formats such as MATLAB (.mat), Excel (.xlsx/.xls), and Word (.doc/.docx) files are not supported and must be converted. For example, convert tabular data to CSV. Some suggested formats for different data uses are listed here.

Database (tabular data)

Required

  • Provide information needed to reuse the data (e.g., data dictionary in Content description)

Recommended

  • Use Comma-Separated Values (CSV) files for small datasets
  • CSVs should be formatted according to the RFC 4180 specification.
  • Structure data following the principles of “tidy data”. For example, each variable should be in a column and each observation (or case) in a row.

Database (waveforms)

Required

  • WFDB compatibility: Store high-temporal-resolution data (ECG, EEG, and similar) in WFDB format or another open-source format such as EDF. Put gain and baseline in the file headers. For detailed guidance on creating signal files, see this MIT-format tutorial and this EDF-format tutorial.
  • RECORDS file: Provide a file named exactly RECORDS in the top-level directory. It lists every WFDB record (without .hea or .dat) or EDF file (with .edf) in your submission, one per line. For records in subdirectories, include the relative path: for example, ECG/subject1_ecg. Example files can be seen here and here.
  • Signal naming convention: Name signals using the PhysioNet standard signal name list. Use upper case where it improves readability (e.g., ABP_Sys is better than abp_sys).
    • If your signals are not in the standard list, add a plain-text file named CALIBRATION with one line for each new signal type, giving its default vertical plotting scale. Examples of this format are at the top of wfdbcal and in wfdbcal(5).
  • Annotations or event locations: Provide annotations and event locations as WFDB annotation files. You can use the MATLAB Toolbox or the Python Toolbox to create annotations. See the command wrann.

Recommended

  • Additional subject data: Provide subject information either at the bottom of the signal header files or in a separate text file. Preferred information includes age, sex or gender, height, weight, medications, and diagnoses.
  • Check for valid signal types (WFDB format): Check your WFDB files for valid signal types with wfdbcheck. This program will attempt to find errors that have occurred previously in both signal and annotation files. It is not comprehensive. Please report any errors it misses, and describe any new signal types you would like added.
  • Visualize and check your waveforms using LightWAVE: If you upload the RECORDS file, you will see a link to view your project's waveforms in LightWAVE. Use it to check your waveforms.

Software

Required

  • Documented installation and usage instructions.
  • A file listing dependencies (a requirements file or similar).

Recommended

  • Unit tests that demonstrate the major features work.
  • Standard style guidelines should be followed where appropriate (for example, PEP 8 for Python).

Model

Required

  • Most machine learning-based projects must be published in a reputable, peer-reviewed journal or conference before they can be considered for full review and publication on PhysioNet. The associated publication should provide a clear description of the methodology, rigorous validation, and appropriate comparisons with relevant baselines.
  • When sharing model weights or other model related files, authors must use safe, non-executable file formats. For example, safetensors is preferred for neural network model weights. PhysioNet does not accept .pkl/pickle files for model sharing because they can execute arbitrary code when loaded and therefore pose a security risk.

Recommended

  • Provide code that loads and runs the model.

Submit Your Project

Click here to create your project.

After You Submit

Editorial Decisions

Once your project is submitted, it will be reviewed by one or more content specialists. Based on this review, you will be provided with an initial editorial decision within 6 weeks of submission. We aim to review subsequent resubmissions within 2 weeks.

Decision Meaning What you do
Accept The editor is mostly satisfied with the content. The submission continues into the copyedit stage. Once a project is accepted, authors will no longer be able to directly edit the project. Review and approve the copyedited version.
Resubmit The content is suitable for publication, but changes are required to make it clearer or more reusable. Changes may include adding required information, restructuring files, and rewording content. Make the changes and resubmit by the deadline (3 months for minor revisions and 6 months for major revisions). Contact your editor to request an extension on this deadline.
Reject The content is not suitable for publication. Read the editor's comments. You may contact the editorial team with questions.

How long the process takes depends on how closely your project follows these guidelines and how quickly you respond to review comments.

Copyedit and publication

After acceptance, the editorial team copyedits your project. You are contacted to review and approve the changes, and the project is published once you approve. After publication the content is fixed.

Updating a published project

To change anything in a published project, authors can find their published project on their project page and then click "new version." They can then make their changes, increase the version number, describe the changes in Release Notes, and submit it for review.