🔥 fhiry - FHIR to pandas dataframe for data analytics, AI and ML

Virtual flattened view of FHIR Bundle / ndjson / FHIR server / BigQuery!

🔥 FHIRy is a python package to facilitate health data analytics and machine learning by converting a folder of FHIR bundles/ndjson from bulk data export into a pandas data frame for analysis. You can import the dataframe into ML packages such as Tensorflow and PyTorch. FHIRy also supports FHIR server search and FHIR tables on BigQuery.

Test this with the synthea sample or the downloaded ndjson from the SMART Bulk data server. Use the 'Discussions' tab above for feature requests.

✨ Checkout this template for Multimodal machine learning in healthcare!

UPDATE 1

Recently added support for LLM based natural language queries of FHIR bundles/ndjson using llama-index. Please install the llm extras as follows. Please be cognizant of the privacy issues with publically hosted LLMs. Any feedback will be highly appreciated. See usage!

pip install fhiry[llm]

See usage.

UPDATE 2

Added support for converting a FHIR Bundle to its textual representation for LLMs. You can also convert individual FHIR resources including Patient, Condition, Observation, Procedure, Medication, AllergyIntolerance and DocumentReference.

from fhiry import FlattenFhir
bundle = json.load(jsonfile)
flatten_fhir = FlattenFhir(bundle)
print(flatten_fhir.flattened)

Installation

Stable

pip install fhiry

Latest dev version

pip install git+https://github.com/dermatologist/fhiry.git

Usage

1. Import FHIR bundles (JSON) from folder to pandas dataframe

import fhiry.parallel as fp
df = fp.process('/path/to/fhir/resources')
print(df.info())

Example source data set: Synthea

Jupyter notebook example: notebooks/synthea.ipynb

2. Import NDJSON from folder to pandas dataframe

import fhiry.parallel as fp
df = fp.ndjson('/path/to/fhir/ndjson/files')
print(df.info())

Example source data set: SMART Bulk Data Server Export

Jupyter notebook example: notebooks/ndjson.ipynb

3. Import FHIR Search results to pandas dataframe

Fetch and import resources from FHIR Search API results to pandas dataframe.

Documentation: fhir-search.md

Example: Import all conditions with a certain code from FHIR Server

Fetch and import all condition resources with Snomed (Codesystem http://snomed.info/sct) Code 39065001 in the FHIR element Condition.code (resource type specific FHIR search parameter code) to a pandas dataframe:

from fhiry.fhirsearch import Fhirsearch

fs = Fhirsearch(fhir_base_url = "http://fhir-server:8080/fhir")

my_fhir_search_parameters = {
    "code": "http://snomed.info/sct|39065001",
}

df = fs.search(resource_type = "Condition", search_parameters = my_fhir_search_parameters)

print(df.info())

4. Import Google BigQuery FHIR dataset

from fhiry.bqsearch import BQsearch
bqs = BQsearch()

df = bqs.search("SELECT * FROM `bigquery-public-data.fhir_synthea.patient` LIMIT 20") # can be a path to .sql file

Filters

Pass a config json to any of the constructors:

• config_json can be a path to a json file.

df = fp.process('/path/to/fhir/resources', config_json='{ "REMOVE": ["resource.text.div"], "RENAME": { "resource.id": "id" }  }')

fs = Fhirsearch(fhir_base_url = "http://fhir-server:8080/fhir", config_json = '{ "REMOVE": ["resource.text.div"], "RENAME": { "resource.id": "id" }  }')

bqs = BQsearch('{ "REMOVE": ["resource.text.div"], "RENAME": { "resource.id": "id" }  }')

Columns

• see df.columns

patientId
fullUrl
resource.resourceType
resource.id
resource.name
resource.telecom
resource.gender
...
...
...

Documentation

Give us a star ⭐️

If you find this project useful, give us a star. It helps others discover the project.

Contributors

• Bell Eapen |
• Markus Mandalka
• PR welcome, please see CONTRIBUTING.md
• using CC