fhir4ds 0.5.1

Production-ready healthcare analytics platform providing 100% SQL-on-FHIR v2.0 compliance with dual database support (DuckDB + PostgreSQL)

FHIR for Data Science (FHIR4DS)

Production-ready healthcare analytics platform providing 100% SQL-on-FHIR v2.0 compliance with dual database support (DuckDB + PostgreSQL).

Current Version: 0.5.0 (Complete FHIRPath Specification Implementation)
Test Compliance: 100% (117/117 tests passing) Database Support: DuckDB 100% + PostgreSQL 100%
User Experience: One-line setup, multi-format export, batch processing

Key Features

• 100% SQL-on-FHIR v2.0 compliance - All 117 official tests passing
• Dual database support - DuckDB and PostgreSQL with identical functionality
• Advanced FHIRPath parsing - Complete parser with 187 choice type mappings
• CTE-based SQL generation - 95% reduction in SQL complexity
• Parallel processing - High-performance data loading and query execution
• Multi-format export - Pandas, JSON, CSV, Excel, Parquet
• Database object creation - Views, tables, schemas from ViewDefinitions

Quick Start

Installation

pip install fhir4ds

Database Setup

from fhir4ds.datastore import QuickConnect

# DuckDB (recommended for analytics)
db = QuickConnect.duckdb("./healthcare_data.db")

# PostgreSQL (enterprise-grade)
db = QuickConnect.postgresql("postgresql://user:pass@host:5432/db")

# Auto-detect from connection string
db = QuickConnect.auto("./local.db")  # → DuckDB

Load Data

# Load FHIR resources with parallel processing
db.load_resources(fhir_resources, parallel=True)

# Load from JSON files (optimized for large files)
db.load_from_json_file("fhir_bundle.json", use_native_json=True)

Execute Queries and Export

# Execute ViewDefinitions with immediate export
df = db.execute_to_dataframe(view_definition)
db.execute_to_excel([query1, query2], "report.xlsx", parallel=True)
db.execute_to_parquet(analytics_query, "dataset.parquet")

# Batch processing with progress monitoring
results = db.execute_batch(queries, parallel=True, show_progress=True)

Create Database Objects

# Create analytics infrastructure
db.create_schema("clinical_analytics")
db.create_view(patient_view, "patient_demographics")
db.create_table(observation_view, "vital_signs_table")

# List created objects
tables = db.list_tables()
views = db.list_views()

For more details, please see the documentation.

Testing

Run the comprehensive test suite:

# Test both DuckDB and PostgreSQL dialects
python tests/run_tests.py --dialect all

# Test specific dialect
python tests/run_tests.py --dialect duckdb
python tests/run_tests.py --dialect postgresql

Test results: 117/117 tests passing on both DuckDB and PostgreSQL.

License

GNU General Public License v3 (GPLv3)

Related Projects

• SQL-on-FHIR Specification - Official specification
• FHIR R4 Specification - FHIR standard
• DuckDB - High-performance analytics database
• PostgreSQL - Enterprise-grade relational database