FHIRPACK (FHIR Python Analysis Conversion Kit) is a general purpose library that simplifies the access, analysis and representation of FHIR and EHR data using PANDAS, an ETL philosophy and a functional syntax. FHIRPACK was designed and developed at the Institute for Artificial Intelligence in Medicine (IKIM) and the Database Systems Research Group of the University of Heidelberg (HDDBS).
The FHIR standard is a promising framework for interacting with healthcare data. However, expressive tools for efficient FHIR server interaction are few and are either too low level, and powerful, or too abstract, and lack features. FHIRPACK provides an easy-to-use and intuitive API that enables effortless access to FHIR data. We strive for a balance between flexibility, usability and feature richness in order to make interacting with FHIR data less painful.
π | |
---|---|
π tutorial | learn about FHIR, FHIRPACK and PANDAS with our example Jupyter Notebooks |
βοΈ email | jayson.salazar@uk-essen.de or salazar@informatik.uni-heidelberg.de |
π€ talk | join the Slack channel for the latest updates and discussions around FHIR, FHIRPACK and EHR in general |
π issues | bugs and feature requests go preferrably on the main GitLab tracker or on GitHub |
π§ dev | learn more about contributing to FHIRPACK or extending its functionality |
π docs | read our documentation on Read the Docs |
π citation | cite FHIRPACK as https://doi.org/10.5281/zenodo.8006589 or visit Zenodo for more details |
You can easily install the latest FHIRPACK release from PyPI or the most current version by cloning this repository. As usual with Python, we strongly recommend using virtual environments such as pipenv, venv or conda.
FHIRPACK requires Python 3.9 or greater as well as libmagic
to work without problems. In case you're using an older Python version, you can use asdf, conda or pyenv to have several Python versions coexist on your system.
libmagic
can be installed with apt-get install libmagic-dev
on Ubuntu or brew install libmagic
on MacOS.
pip install fhirpack
alternatively use pipenv:
pipenv --python 3.9 install fhirpack
git clone https://gitlab.com/fhirpack/main.git
mv main fhirpack
pip install -e fhirpack
alternatively use pipenv:
git clone https://gitlab.com/fhirpack/main.git
mv main fhirpack
pipenv install -e fhirpack
You can use pip install --upgrade fhirpack
or if you're using pipenv pipenv update fhirpack
to upgrade to the latest release of FHIRPACK once you've already installed it.
To set up a server configuration, create an .env
file in the directory where you'll work with FHIRPACK and specify the settings as can be seen in .env.example.
Alternatively, copy, rename and modify .env.example
according to your needs.
FHIRPACK is based on the ETL paradigm, and as all functions available to you can be classified as Extractors, Transformers or Loaders. In the following examples we quickly show how this works, but remember to have a look at the examplary Jupyter Notebooks and the API reference
In this example we extract all the conditions for a patient with the ID: 43fb1577-3455-41cf-9a07-c45aa5c0219e
from the public FHIR-server with the Base-URL: http://hapi.fhir.org/baseR4.
# import FHIRPACK
import fhirpack as fp
# instantiate a connected PACK for the specified FHIR API base
pack = fp.PACK("http://hapi.fhir.org/baseR5")
# retrieve a list of Patients of length one by ID
patient = pack.getPatients(["43fb1577-3455-41cf-9a07-c45aa5c0219e"])
# retrieve all respective conditions for said patients
condition = patient.getConditions().explode()
# gather and display the FHIR elements with the specified paths from the conditions
condition.gatherSimplePaths(["id", "code.coding.code", "code.coding.display", "onsetDateTime" ])
id | code.coding.code | code.coding.display | onsetDateTime | |
---|---|---|---|---|
0 | 2a65f2a4-1a8d-46d9-a5f9-3af95a5d99bd |
[267036007] |
[Dyspnea (finding)] |
2020-02-23T12:07:58-06:00 |
1 | c9f11f99-796c-4c34-9a8e-246f1faa0039 |
[840544004] |
[Suspected COVID-19] |
2020-02-23T12:07:58-06:00 |
2 | d5c30da3-546c-486c-bdb4-ff8f1b62a553 |
[386661006] |
[Fever (finding)] |
2020-02-23T12:07:58-06:00 |
3 | a9c2b72d-b6de-4544-95d8-16246786fb5b |
[49727002] |
[Cough (finding)] |
2020-02-23T12:07:58-06:00 |
4 | dd0b2c03-75fe-4d2e-9d49-45c543f5c825 |
[840539006] |
[COVID-19] |
2020-02-23T13:26:58-06:00 |
# import FHIRPACK
import fhirpack as fp
# instantiate a connected PACK for the specified FHIR API base
pack = fp.PACK("http://hapi.fhir.org/baseR5")
# retrieve all conditions containing the term sepsis
conditions = pack.getConditions(searchParams={"_content": "sepsis"})
# get the respective patients
patients = conditions.getPatients().explode()
# display the specified FHIR elements of the patients
patients.gatherSimplePaths([
"name.given",
"name.family",
"telecom.value",
"address.country",
"address.city",
"birthDate"
])
name.given | name.family | address.country | address.city | birthDate | |
---|---|---|---|---|---|
0 | [[Herbert]] |
[Hoover] |
None |
[Everytown |
1990-07-04 |
1 | [[Aaron697]] |
[Stiedemann542] |
[US] |
[Westford] |
1946-03-29 |
2 | [[Aaron697]] |
[Stiedemann542] |
[US] |
[Westford] |
1946-03-29 |
3 | [[Aaron697]] |
[Stiedemann542] |
[US] |
[Westford] |
1946-03-29 |
4 | [[Aaron697]] |
[Stiedemann542] |
[US] |
[Westford] |
1946-03-29 |
5 | [[Aaron697]] |
[Stiedemann542] |
[US] |
[Westford] |
1946-03-29 |
6 | [[Charlesetta336], [Charlesetta336]] |
[Kihn564, Pouros728] |
[US] |
[Falmouth] |
1943-12-17 |
7 | [[Hiram237]] |
[Kertzmann286] |
[US] |
[Fall River] |
1999-06-07 |
11 | [[Tim]] |
[Shabad] |
None |
None |
1980-01-01 |
βΉοΈ For more examples and a deep-dive into FHIRPACK, please take a look at the example jupyter notebooks.
FHIRPACK also provides a CLI for easy and quick data exploration.
The CLI can be invoked by using python -m fhirpack.cli
or fp
once FHIRPACK has been installed.
> fp --help
Usage: fp [OPTIONS]
The pack for your FHIR server.
Options:
--version Show the version and exit.
-s, --source TEXT URL of the FHIR server or path to json files.
-e, --environment TEXT Path to Dotenv file containing configurations.
-d, --destination TEXT Location of the output.
-o, --operation TEXT Operations to retrieve data.
-p, --params TEXT Provide additional search parameters.
-v, --verbose Print results in verbose format.
-h, --help Show this message and exit.
CLI usage is analogous to the general fhirpack
dataflow.
Python | CLI |
---|---|
pack.getPatients(["1"]) |
fp -o "getPatients 1" |
pack.getPatients(["1", "181", "525"]) |
fp -o "getPatients 1, 181, 525" |
PACK(envFile=".env.example").getPatients(["1"]) |
fp -e .env.example -o "getPatients 1" |
pack.getPatients(searchParams={}).gatherSimplePaths(["name.family"]) |
fp -o "getPatients" -p all -o "gatherSimplePaths name.family" |
pack.getPatients(searchParams={"family":"Koepp"}) |
fp -o "getPatients" -p "family = Koepp" |
Please report any bugs you find on our main GitLab Tracker or here as well. If you want to contribute a fix or feature, you're welcomed to create a pull request from your fork/branch or create a merge request on the main GitLab repository according to our contribution guidelines.
If you're using FHIRPACK, feel free to reach out over Twitter, Slack or e-mail
FHIRPACK is released under the MIT, see LICENSE.txt for more information.
For scientific work and formal citations, please use:
@software{fhirpack,
author = {Jayson Salazar Rodriguez},
title = {FHIRPACK (FHIR Python Analysis Client and Kit)},
month = oct,
year = 2022,
publisher = {Zenodo},
version = {v0.0.10b0},
doi = {10.5281/zenodo.8006589},
url = {https://doi.org/10.5281/zenodo.8006589}
}