AI-assisted population-PK analysis
Run an independent pop-PK analysis and compare it with the applicant's results, entirely from GitHub. Demonstrator for IncreaseNET Task 7.7 · synthetic data only · tool 0.1.0
2analysis requests
5runs
1rejected as out of scope
Start a new analysis
Only fictitious identifiers here: this demonstrator runs in a public repository. Identifiers such as XYZ-123 are pseudonymised before the language model sees the text, and the data never leave the analysis engine.
This opens a pre-filled GitHub issue: click Create. The tool then replies in the issue thread within about two minutes. Running analyses requires being a collaborator of the repository.
How it works
Request. Your text becomes an issue; the language model turns it into a structured spec. It writes no code and sees no data.
Clarify. Missing methodology (e.g. BLQ handling) is asked, never guessed. Reply in the thread.
Confirm. Review the proposed spec and comment /confirm. Anything outside the template catalogue is rejected.
Run. A versioned nlmixr2 template runs in a pinned container with the network disabled.
Review. The report flags differences with the applicant without concluding. Refine with /refine …, check with /replay.
Analysis requests
| Issue | Request | Status | Runs | Last run | Updated |
|---|---|---|---|---|---|
| #2 | Reproduce the applicant's model for XYZ-123: two compartments, firs… | awaiting answer | 0 | – | 2026-10-08T14:34:24Z |
| #1 | Reproduce the applicant's model for XYZ-123 (end-to-end test) | completed | 5 | 141300-3341 |
2026-10-08T14:14:30Z |
Explore
Run history (assessor requests) Reference demo run by CI Source code Demo walkthrough
Generated 2026-10-08T14:34:27Z.