How common-longitudinal-icu-data-format/pyclif works
Project type and purpose
• pyCLIF is an open-source Python library (not a web or full-stack app) designed to load, validate, transform, and analyze CLIF (Common Longitudinal ICU Data Format) data.
• Its primary users are data engineers, clinical data scientists, and developers who need a standardized interface over ICU-related clinical tables.
File structure
Top-level directories and files:
• src/pyclif/ ← main package
– clif.py ← core orchestrator class
– data/ ← demo datasets + loader
– mCIDE/ ← JSON schemas for each CLIF table
– tables/ ← implementations of the nine CLIF table classes
– utils/ ← I/O, validation, timezone handling, dataset transforms
• config/config.json ← default configuration for loader or validation
• CQODE/ ← R-based demo functions (separate client code)
• examples/ ← Jupyter notebooks demonstrating usage
• tests/ ← pytest suite covering core, tables, utils
• pyproject.toml ← build and dependency definitions
• README.md ← high-level overview, usage, status
README for additional insights
• Confirms nine CLIF-2.0.0 tables, validation against mCIDE schemas, timezone handling, filtering, chunked processing, and a core CLIF class API.
• Mentions key dependencies: pandas, duckdb, pyarrow, pydantic, pytz.
• Demonstrates two usage patterns: high-level CLIF orchestrator vs. per-table direct loading.