This repository is a practical guide to CLIFing—the process of transforming local clinical EHR data into a CLIF.
| Resource | Description |
|---|---|
| Epic UserWeb: CLIF SQL Queries | SQL queries for Epic Caboodle/Clarity to extract CLIF tables (requires Epic UserWeb access) |
| CLIF-MIMIC | Reference ETL implementation converting MIMIC-IV → CLIF |
| CLIF Data Dictionary | Official schema definitions and mCIDE vocabularies |
| CLIF 101 | Beginner's guide to working with CLIF data |
| clifpy | Python package for validation and analysis |
| CLIF Lighthouse | Interactive QC dashboard |
Different institutions take different paths to build their CLIF instances. There is no one-size-fits-all solution, as source EHR systems, data availability, and local constraints vary widely.
The goal of this repo is to document how we built our CLIF pipeline, starting from a local EHR and progressing step by step to a CLIF-ready dataset, and to share lessons learned along the way. (we try to share as much as we can)
- The overall pathway from EHR → CLIF
- Design decisions and trade-offs we made
- Mapping strategies for diagnoses, procedures, labs, and ICU concepts
- Common challenges encountered during CLIF implementation
- Practical examples and scripts used in our pipeline
This is not intended to be a strict template, but rather a reference and starting point for other teams building their own CLIF instances.
- A plug-and-play CLIF solution
- A definitive or “correct” way to build CLIF
- A replacement for official CLIF documentation
Instead, it reflects one real-world implementation and is meant to complement existing CLIF resources.
- Researchers and engineers working with ICU or clinical EHR data
- Institutions planning to adopt or pilot CLIF
- Teams looking for practical examples beyond high-level specifications
- Review the documented pipeline to understand the full CLIFing process
- Adapt relevant steps to your local data and infrastructure
- Use examples as guidance, not strict requirements
We encourage sites to share their ETL approaches! To contribute:
- Create a folder with your site/institution name (e.g.,
RUSH/,UCMC/) - Add your data transformation scripts (censored of any PHI or proprietary details)
- Include a brief README explaining your EHR source and approach
- Submit a PR
The more examples we collect, the easier it becomes for new sites to onboard.
- Join the #clif-code-ecosystem channel in the CLIF Consortium Slack
- Check the Epic UserWeb thread for Epic-specific discussions
- Review CLIF-MIMIC for a complete reference implementation