Google's tools for working with FHIR data in Python. This includes:
- Support for converting FHIR data to and from an efficient Protocol Buffer-based format.
- Support for creating and analyzing views over large FHIR datasets. See the Google FHIR Views documentation and notebook examples for details.
This is not an officially supported Google product.
Note: Requires Python >= 3.8.
These libraries are installed via pip.
Users interested in FHIR Views and the underlying libraries can simply run the following to install the views library, BigQuery, Spark and FHIR R4 dependencies:
pip install google-fhir-views[r4,bigquery,spark]
Users who only need the BigQuery or Spark runners can run:
pip install google-fhir-views[r4,bigquery]
or
pip install google-fhir-views[r4,spark]
respectively, to reduce the installation size.
Note: If installing for use in a Jupyter notebook, it's best pip install ... before starting the notebook kernel to avoid dependency version issues.
Users looking for only the underlying FHIR Protocol Buffer support can
run pip install google-fhir-r4 to retrieve only that and its dependencies.
This can be installed locally, directly from source by running the following commands in this directory. As always, doing so within a Python virtual environment is recommended.
This library generates Protocol Buffers for FHIR resources, so the protoc executable must be available. This can be done on Linux by running:
apt install protobuf-compiler
protoc --version # Ensure version 3+
Or on MacOS with Homebrew:
brew install protobuf
protoc --version # Ensure version 3+
Windows users can download protoc releases here.
Once protoc is available, the fhir-py libraries can be installed from source by running the following in the fhir-py directory:
pip install ./google-fhir-core[bigquery,spark]
pip install ./google-fhir-r4
pip install ./google-fhir-views[r4,bigquery,spark]
See the Google FHIR Views documentation for details on use.
Due to the nature of the initial commit squashing internal contributor history, we would like to recognize some of those who contributed to the initial commit work: Ryan Brush (@rbrush), Cameron Tew (@cam2337), Ose Umolu (@luid101), Walt Askew (@waltaskew), Nick George (@nickgeorge), Wilson Sun (@wilsonssun) Lisa Yin (@lisayin), Suyash Kumar (@suyashkumar), and other Googlers.
In addition, contributors who committed after the initial squash commit can be found in the GitHub contributors tab.
Thank you to all contributors!
FHIR® is the registered trademark of HL7 and is used with the permission of HL7. Use of the FHIR trademark does not constitute endorsement of this product by HL7.