Output schema

run_pipeline.py writes dashboard_data.json, one JSON object holding the whole bundle. This page lists its keys and the fields of each record. The field lists were taken from a run on the invented ALS cohort on 9 November 2026; run examples/make_synthetic_cohort.py and run_pipeline.py to reproduce them. A table that is empty in your data may have no rows at all.

Top-level keys

{
  "metadata": {"generated_at": "...", "pipeline": "Registry Forge", "total_patients": 58, "warnings": []},
  "patients": [], "documents": [], "notes": [],
  "encounters": [], "problems": [], "medications": [], "procedures": [],
  "labs": [], "vitals": [], "labs_vitals": [],
  "allergies": [], "immunizations": [], "careplans": [], "goals": [],
  "diagnostic_reports": [], "document_references": []
}

metadata.warnings lists identity and input problems found during the run, such as name and birth date matches left unmerged and attachments that were not delivered.

Fields shared by clinical records

Field	Contents
patient_id	Joins to patients[].patient_id
code, code_system, display_name	The code chosen for display (see Stage 3) and its system and name
all_codings	Every coding on the source element, each with code, system_name, system_oid and display
effective_date, effective_datetime	The record's main date
status	Status from the source, where present
source	fhir or ccda
source_file	The file the record came from

Tables

Table	Fields beyond the shared ones
patients	first_name, last_name, dob, gender, race, ethnicity, mrn, source_patient_id, source_patient_ids, source_system, deceased, deceased_date_time, num_documents (no shared clinical fields)
documents	document_uuid, document_title, document_date, source_format (ccda_xml, pdf, rtf, html_fragment, unknown), source_type (rtf_note for RTF; fhir_attachment for text from FHIR resources; otherwise as source_format), fhir_resource_type, plain_text (up to 1,000,000 characters; longer text is cut and text_truncated is set), source_file
notes	document_id, document_date, section_title (for example "Problems" or "Plan of Care"), narrative_text, character_count; one row per C-CDA section
problems	clinical_status, onset_datetime
medications	fhir_resource_type (MedicationRequest, MedicationDispense, MedicationAdministration or MedicationStatement), medication_subtype, authored_on, start_date
labs, vitals, labs_vitals	value, value_text, unit, category
procedures	performed_datetime
encounters	type, visit_type, visit_type_display, start_date, end_date
diagnostic_reports	conclusion
document_references	type_display
allergies	allergen, allergen_code, allergen_system, recorded_date, clinical_status
immunizations	vaccine, vaccine_code, vaccine_system, occurrence_datetime, recorded_date
careplans, goals	Title or description, status and date from the source

Aliases such as start_date and authored_on hold the same value as the canonical field; see Stage 4.

Example record

A medication from the single-patient sample data:

{
  "patient_id": "demo-patient-001",
  "code": "35623",
  "code_system": "RxNorm",
  "display_name": "Riluzole",
  "all_codings": [
    {"code": "35623", "system_oid": "http://www.nlm.nih.gov/research/umls/rxnorm",
     "system_name": "RxNorm", "display": "Riluzole"}
  ],
  "effective_date": "2023-06-01",
  "effective_datetime": "2023-06-01",
  "authored_on": "2023-06-01",
  "start_date": "2023-06-01",
  "status": "active",
  "source": "fhir",
  "source_file": "demo-fhir-001.json",
  "fhir_resource_type": "MedicationRequest",
  "medication_subtype": "MedicationRequest"
}

RxNorm 35623 is the riluzole ingredient, checked against RxNorm in November 2026.