This repository is a practical guide to CLIFing—the process of transforming local clinical EHR data into a CLIF.
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