Example FHIR to TranSMART loader

This package contains a mapper that reads data from HL7 FHIR (STU 3 or R4) resources and translates them to the data model of the TranSMART platform, an open source data sharing and analytics platform for translational biomedical research.

It also provides a utility that applies the mapper and writes the translated data to tab-separated files that can be loaded into a TranSMART database using the transmart-copy tool.

The FHIR reader is based on the fhirclient package, writing to TranSMART is based on transmart-loader.

⚠️ Note: this is a very preliminary version, still under development. Issues can be reported at https://github.com/thehyve/python_fhir2transmart/issues.

Installation

The package requires Python 3.6.

To install fhir2transmart, do:

pip install fhir2transmart

Or from source:

git clone https://github.com/thehyve/python_fhir2transmart.git
cd python_fhir2transmart
pip install .

Run tests (including coverage) with:

python setup.py test

Usage

Read input from a JSON file input.json and write the output in transmart-copy format to /path/to/output. The output directory should be empty of not existing (then it will be created).

# Translate one json file
fhir2transmart [--with-ontology] input.json /path/to/output
# Translate all json files in a directory
fhir2transmart [--with-ontology] input_dir /path/to/output

With the --with-ontology option, concepts and ontology nodes are created based on the input data. Do not use this option when ontologies have already been loaded separately.

Example data is available at MITRE SyntheticMass. Instructions:

# Download 1K Sample Synthetic Patient Records, FHIR STU3 : 20MB
wget https://syntheticmass.mitre.org/downloads/2017_11_06/synthea_sample_data_fhir_stu3_nov2017.zip
# unzip creates a directory 'fhir' containing 282MB of json files
unzip synthea_sample_data_fhir_stu3_nov2017.zip
# create an output directory
mkdir output
# apply the mapping
fhir2transmart fhir output

This generates the directories i2b2metadata and i2b2demodata in the output directory. The generated data can be loaded using transmart-copy:

# Download transmart-copy:
curl -f -L https://repo.thehyve.nl/service/local/repositories/releases/content/org/transmartproject/transmart-copy/17.1-HYVE-5.9-RC3/transmart-copy-17.1-HYVE-5.9-RC3.jar -o transmart-copy.jar
# Load data
PGUSER=tm_cz PGPASSWORD=tm_cz java -jar transmart-copy.jar -d output

Mapping

The following mapping table shows how FHIR resources are mapped to the TranSMART data model.

FIHR	TranSMART
Resource type attribute	Class attribute concept modifier
============= =================	============== ============== ============ =========
Patient identifier	PatientMapping identifier
Patient gender	Patient sex
Patient gender	Observation value Gender
Patient birthDate	Observation value BirthDate
Patient deceased	Observation value Deceased
Patient deceasedDate	Observation value DeceasedDate
------------- -----------------	-------------- -------------- ------------ ---------
Condition subject	Observation patient
Condition code	Observation conceptCode
Condition onsetDateTime	Observation startDate
Condition abatementDateTime	Observation endDate
Condition recordedDate Condition category	Observation
------------- -----------------	-------------- -------------- ------------ ---------
Encounter identifier	Visit
Encounter period.start	Visit startDate
Encounter period.end	Visit endDate
Encounter status	Visit activeStatusCd
Encounter class	Visit inoutCd
Encounter hospitalization	Visit locationCd

Acknowledgement

This project was funded by the German Ministry of Education and Research (BMBF) as part of the project DIFUTURE - Data Integration for Future Medicine within the German Medical Informatics Initiative (grant no. 01ZZ1804D).

License

Copyright (c) 2019 The Hyve B.V.

The FHIR to TranSMART loader is licensed under the MIT License. See the file LICENSE.