Hi all, this is my first tool post here, following up on the intro I just posted in the General thread.
I’ve been building an open-source R package called rwevalidate and would really value feedback from this community.
It validates whether a patient cohort is fit for a real-world evidence study on OMOP CDM data. One validate_cohort() call produces a structured HTML + JSON report covering cohort attrition, prior-observation coverage, demographics, and temporal data density, with each check mapped to the FDA Relevance + Reliability framework, HARPER, and RECORD-PE.
The motivation: DataQualityDashboard and Achilles do a great job characterizing a database, but I kept needing to answer a narrower, study-specific question, is this particular cohort fit for this particular analysis, in the terms a regulator uses? rwevalidate is meant to complement those tools, not replace them. It is deliberately lightweight and independently installable (no HADES dependencies), and it works offline against a built-in synthetic CDM so you can see the report without a database.
Installable from GitHub today; a CRAN submission is currently under review:
remotes::install_github("tavakohr/rwevalidate")
GitHub (with a vignette and example report): GitHub - tavakohr/rwevalidate: Validate OMOP CDM patient cohorts for real-world-evidence studies (FDA RWE + HARPER aligned HTML/JSON report) · GitHub
I’d love feedback on the validation checks, the threshold defaults, and whether the regulatory mappings are useful in your work. I’m also looking for collaborators, and would welcome pointers on which workgroup this best fits. Happy to give a short demo on a community call if there’s interest.
Thanks!