R Package Assessment: Lessons from Pharma
summary
GSK describes a concrete, audit-ready checklist centered on five domains of trustworthiness that evaluates R packages for regulatory use: downloads and trajectory to gauge real-world vetting, authorship/prior work to assess statistical expertise, documentation of inputs/outputs and edge cases, and SDLC signals like protected branches and CI plus release cadence; testing is judged by unit tests and code coverage and comparisons to external “source of truth.” Guidance includes thresholds for submission vs exploratory use and when to add extra QC.
AI-written summary · about
chapters (10)
- 0:00Introduction and context
- 2:13Defining trust in pharma
- 5:09The five domains of trust
- 5:37Community domain
- 7:19Authorship domain
- 8:34Documentation domain
- 9:53Development life cycle
- 12:07Testing domain
- 14:21Risk-based implementation
- 16:48Holistic perspective
related talks
11mField Guide to Writing Your First R Package
04mThe Expanse - Navigating the R Package Universe
19mAutomated Techniques for Measuring Trustworthiness of Open Source Code and Communities
28mFinding Dependable Go Packages
19mIt's a Great Time to be an R Package Developer!
18mHow non-programmers can make sparks fly when using testthat during package development
26mTheory of programming: packaging principles
14mBeyond RAG: Building Reliable AI Systems for Privacy Assessments