Downloads¶
Every file on this page is built from the repository each time the site is built, so what you download matches the documentation you are reading.
Everything in one file¶
The zip holds the pipeline and add-on modules, the record viewer, the Databricks export, the single-patient sample export, the invented-cohort generator, the run notebook, the tests, requirements.txt, the license and citation files. Unzip it and follow the Quickstart, or see Where to run it for Jupyter, Google Colab and AWS.
Run notebook¶
Download run_registry_forge.ipynb
One notebook that checks the environment, finds or fetches the pipeline, installs missing packages, builds the invented cohort (or reads your working directory), runs everything and shows where each output went. It works in JupyterLab, Jupyter Notebook, JupyterHub, Google Colab and Amazon SageMaker AI.
Pipeline files one at a time¶
Core pipeline, add-on modules and record viewer, each with what it does
| File | What it does |
|---|---|
run_all.py |
Runs the core pipeline and every add-on module in order and sorts the outputs into identified, restricted and counts-only folders |
run_pipeline.py |
The core pipeline, Stages 1 to 7, with note extraction as Stage 7c. Paths are set with --base-dir |
dashboard.html |
Browser record viewer. Open it locally and load dashboard_data.json; identifiers are hidden by default |
note_extraction.py |
ALS note patterns (ALSFRS-R, ECAS, El Escorial, onset, family history, genes, FVC, treatment milestones) and the extraction itself. See Note extraction |
note_dashboard.py |
Note extraction summary, and an optional restricted review file. See Note extraction dashboard |
dashboard_common.py |
Shared layout and suppression rules for the device and exposure dashboards |
cohort_eda.py |
Cohort summary report. See Cohort EDA report |
drug_repurposing.py |
Drug repurposing cohort assembly, summary, report and CURE ID review worksheet. See Drug repurposing |
device_extraction.py |
Devices and equipment from HCPCS Level II, SNOMED CT and CPT codes and from text. See Device and equipment extraction |
device_dashboard.py |
Device summary, and an optional restricted review file |
exposure_extraction.py |
Environmental, occupational and toxic exposures from codes and text, with ECTO categories. See Exposure extraction |
geospatial_extraction.py |
Optional local address candidates, source provenance and a human-reviewed restricted queue. Added 13 November 2026; run separately. See Location extraction |
exposure_dashboard.py |
Exposure summary, and an optional restricted review file |
omop_etl.py |
OMOP CDM v5.4 tables from an Athena vocabulary download. See OMOP ETL |
phenopackets_etl.py |
GA4GH Phenopacket v2 export, de-identified by default. See Phenopackets ETL |
mondo_omop_bridge.py |
Cohort code lists from a Mondo disease term, with rare disease subset flags. See Mondo-OMOP bridge |
test_patients.txt |
Template for exclusion rules. See Stage 6 |
Databricks data extraction¶
These run inside Databricks (or any Spark SQL engine) and produce the chunked CSV inputs the pipeline reads. See Data extraction (Databricks) for instructions.
| File | What it does |
|---|---|
databricks_export.py |
PySpark notebook with a configuration block, validation and consolidation into three CSV files |
databricks_export.sql |
Pure SQL alternative for SQL-only Databricks notebooks |
Templates¶
| File | Used for |
|---|---|
vital_status.csv |
Deaths and last confirmed alive dates from sources outside the records, and the data cut-off |
test_patients.txt |
Test records to remove |
synthetic_geospatial_bundle.json |
Fictional notes for practicing the optional location extractor |
Sample data¶
A single invented ALS patient in the export layout, with records in every category, that you can run the core pipeline against without any real EHR access. See Sample data for what is in each file.
| File | What it is |
|---|---|
sample_ccda.xml |
C-CDA continuity-of-care document with every main section populated |
sample_fhir_bundle.json |
FHIR R4 Bundle covering every extracted resource type |
uuid_mapping.csv |
Document-to-patient mapping with demographics |
ccda_chunks.csv |
The C-CDA as the chunked CSV the pipeline reads |
fhir_chunks.csv |
The FHIR bundle as the chunked CSV the pipeline reads |
patient_master.csv |
The core pipeline's patient master CSV for the sample, built at each site build. See Patient master CSV |
note_extractions.csv |
The note extraction candidates for the sample, built at each site build |
Invented ALS cohort¶
Download synthetic-als-inputs.zip
Invented participants (Jane and John Doe shells, IDs SYN-NNNN) with ALS, primary lateral sclerosis, progressive bulbar palsy or unspecified motor neuron disease, laid out like the Databricks export: CCDA and FHIR data/ccda_chunks.csv (C-CDA XML, RTF and HTML clinic notes, genetic reports as PDF or text, hospice summaries), CCDA and FHIR data/fhir_chunks.csv (FHIR bundles from one or two invented health systems, with clinic notes and EMG reports attached), uuid_mapping.csv, vital_status.csv and a README. It can be passed straight to run_all.py as the working directory. The generator, make_synthetic_cohort.py, is deterministic: the same version always writes the same files.