This is an effort to compile a repository of the clinical characteristics of patients who have taken a COVID-19 test. By sharing our schema and data, we hope that we can 1) accelerate information sharing among frontline healthcare providers and 2) facilitate studies on COVID-19 signs, symptoms, stages, and care plans.
The repository is maintained as CSVs and in Google Sheets and is compliant with HIPAA Privacy Rule's De-Identification Standard. Details about each field are available in the associated data dictionary.
- Each row contains the clinical characteristics of a patient who has taken a COVID-19 test.
- Each batch is published as a separate CSV file (and Google Sheets tab).
- A new batch is compiled and published weekly, including data from Carbon Health and Braid Health.
- Each filename is prefixed with the date (mm-dd) the query was run, which matches with the
date_publishedfield. - Each batch contains a week's worth of test results, with the most recent date being
date_published - 1. The first batch—prefixed with 04-07—contains a month's worth of data starting from 03-07.
- Website: Carbon Health
- Twitter: @CarbonHealth
- Email: covidclinicaldata@carbonhealth.com
- Notes:
- Carbon Health began COVID-19 testing with the SARS-CoV-2 RNA RT-PCR test on 03-04-20.
- The data includes clinical characteristics in addition to radiological and laboratory findings. It does not include treatment plans, complications, and clinical outcomes, which is collected at inpatient facilities.
- The data includes both positive- and negative-tested patient characteristics. These include the characteristics of symptomatic patients, those in professions with a high risk of exposure, and/or those who may have been exposed through contact with a known infected person.
- Clinician-assessed symptoms are sparse for data published on 04-07 due to some criteria having been added later.
- A patient's reported age differs from their actual age by a reasonable randomized amount to protect their privacy.
 |
 |
|---|---|
| Carbon Health Logo | Data Dictionary |
- Website: Braid Health
- Twitter: @BraidHealth
- Email: vivian@braid.health and k@braid.health
- Notes:
- The data is merged with Carbon Health clinical fields and includes findings, clinician impressions, and links to chest x-rays.
- The links direct to the Braid Health website. The website UI allows for closer inspection by researchers and radiologists.
- The images can be downloaded for image processing and classification studies.
- A patient’s reported age differs from their actual age by a reasonable randomized amount to protect their privacy.
 |
 |
|---|---|
| Braid Health Logo | Sample Chest X-ray |
Special thanks to Kevin Quennesson, Nigam Shah, Andrew Therriault, and Andrew Pikul for their support of this effort and for their feedback.
To ensure this data is representative of cases with varying severity levels and symptoms, we are requesting data from outpatient test centers and inpatient healthcare facilities which are treating COVID-19. Please use the templates below and email the data to covidclinicaldata@carbonhealth.com.
Details about the fields are available in the data dictionary.
Outpatient test centers and clinics can contribute their data using the outpatient template.
Inpatient healthcare providers can contribute additional columns for treatment plans, complications, and clinical outcomes using the inpatient template.
Please share any studies on this data via email or a pull request. You can use the format below to cite the data repository in your studies.
@dataset{2020covidclinicaldata,
author = {Carbon Health and Braid Health},
title = {Coronavirus Disease 2019 (COVID-19) Clinical Data Repository},
howpublished = {Accessed from \url{https://covidclinicaldata.org/.}},
year = 2020,
version = {3.0}
}
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
