OMOPy

Pythonic, type-safe interface for OMOP CDM databases.

OMOPy is a single Python package that reimplements the DARWIN EU R package ecosystem for working with OMOP Common Data Model databases. It provides lazy database access via Ibis, type-safe data structures via Pydantic and Polars, and a clean Pythonic API with full type hints.

Background

The DARWIN EU Coordination Centre develops and maintains an ecosystem of R packages for observational health research using the OMOP CDM. These include CDMConnector, PatientProfiles, CohortCharacteristics, IncidencePrevalence, DrugUtilisation, CohortSurvival, TreatmentPatterns, and others.

OMOPy consolidates these ~17 R packages into a single Python library, bringing the DARWIN EU analytical toolkit to the Python data science ecosystem. The package preserves the conceptual model and analytical capabilities of the R packages while providing a Pythonic API that follows Python conventions and leverages modern Python tooling.

Features

• Single package — one pip install omopy replaces 17 R packages
• Lazy by default — Ibis constructs SQL queries; nothing executes until you call .collect()
• Type-safe — Pydantic models with frozen immutability; full type annotations throughout
• Pythonic — snake_case, context managers, keyword arguments, no R idioms
• Database-agnostic — DuckDB, PostgreSQL, SQL Server, Snowflake, BigQuery, and more via Ibis backends

Modules

Module	R Equivalent	Description
omopy.generics	omopgenerics	Core type system: CDM references, tables, codelists, summarised results
omopy.connector	CDMConnector	Database connections, CDM loading, cohort generation, CIRCE engine
omopy.profiles	PatientProfiles	Patient-level enrichment: demographics, intersections, death
omopy.codelist	CodelistGenerator	Vocabulary search, hierarchy traversal, codelist operations
omopy.vis	visOmopResults	Format, tabulate, and plot summarised results
omopy.characteristics	CohortCharacteristics	Cohort characterization: summarise, tabulate, plot
omopy.incidence	IncidencePrevalence	Incidence rates and prevalence proportions
omopy.drug	DrugUtilisation	Drug cohort generation, utilisation metrics, dose analysis
omopy.survival	CohortSurvival	Kaplan-Meier survival, competing risks, survival plots
omopy.treatment	TreatmentPatterns	Treatment pathway analysis, Sankey and sunburst plots
omopy.drug_diagnostics	DrugExposureDiagnostics	Drug exposure quality checks and diagnostics
omopy.pregnancy	PregnancyIdentifier	Pregnancy episode identification (HIPPS algorithm)
omopy.testing	TestGenerator	Test data generation for OMOP CDM studies

Installation

pip install omopy

Or with uv:

uv add omopy

Requirements

• Python >= 3.14
• A database with OMOP CDM v5.3 or v5.4 tables

Optional database backends

pip install omopy[postgres]    # PostgreSQL via psycopg
pip install omopy[mssql]       # SQL Server via pyodbc
pip install omopy[snowflake]   # Snowflake
pip install omopy[bigquery]    # Google BigQuery
pip install omopy[all]         # All backends

Quick Start

from omopy.connector import cdm_from_con, generate_concept_cohort_set
from omopy.generics import Codelist

# Connect to a DuckDB OMOP CDM database
cdm = cdm_from_con("path/to/omop.duckdb", cdm_schema="main")

# Define a concept-based cohort
codelist = Codelist({"hypertension": [320128]})
cdm = generate_concept_cohort_set(cdm, codelist, name="hypertension")

# Enrich with demographics
from omopy.profiles import add_demographics
enriched = add_demographics(cdm["hypertension"], cdm)

# Collect to a Polars DataFrame
df = enriched.collect()
print(df)

# Characterise the cohort
from omopy.characteristics import summarise_characteristics, table_characteristics

result = summarise_characteristics(cdm["hypertension"])
table_characteristics(result, type="gt")

# Estimate incidence
from omopy.incidence import (
    generate_denominator_cohort_set,
    estimate_incidence,
    plot_incidence,
)

cdm = generate_denominator_cohort_set(cdm, name="denominator")
inc = estimate_incidence(cdm, "denominator", "hypertension", interval="years")
plot_incidence(inc)

Notebooks

The notebooks/ directory contains 12 fully executable Jupyter notebooks demonstrating every major capability:

Notebook	Topic
01_getting_started	CDM connection, tables, snapshot, subsetting
02_codelist_generation	Vocabulary search, hierarchy, codelist operations
03_cohort_generation	Concept-based and CIRCE/JSON cohort generation
04_cohort_characteristics	Summarise, tabulate, and plot cohort characteristics
05_patient_profiles	Demographics, intersects, categories, enrichment
06_incidence_prevalence	Denominator generation, incidence and prevalence
07_drug_utilisation	Drug cohorts, utilisation metrics, indication
08_cohort_survival	Single-event and competing-risk survival analysis
09_treatment_patterns	Treatment pathways, Sankey and sunburst plots
10_drug_exposure_diagnostics	Drug exposure quality checks
11_pregnancy_analysis	HIPPS pregnancy identification algorithm
12_visualisation_styling	Table and plot formatting, styles

Documentation

Full API documentation is available at darwin-eu.github.io/omopy.

Development

# Clone and install
git clone https://github.com/darwin-eu/omopy.git
cd omopy
uv sync --all-extras --dev

# Run tests (1619 tests)
uv run pytest

# Run linting
uv run ruff check src/ tests/

# Build documentation
uv run python docs/_build.py build --strict

# Install pre-commit hooks
uv run pre-commit install

See CONTRIBUTING.md for the full development guide, including the code style guide, type hint requirements, and pull request process.

About DARWIN EU

DARWIN EU (Data Analysis and Real-World Interrogation Network) is a federated network of data, expertise, and services for generating reliable evidence on the real-world safety and effectiveness of medicines. It is coordinated by the DARWIN EU Coordination Centre and supports regulatory decision-making by the European Medicines Agency (EMA).

OMOPy builds on the analytical methods and tooling developed by the DARWIN EU community and the broader OHDSI (Observational Health Data Sciences and Informatics) network. The OMOP Common Data Model provides the standardised data structure that underpins all analyses.

License

This project is licensed under the MIT License. See LICENSE for details.