From the concentration CSV to an eCTD-ready data package
Pharmacokinetic and bioequivalence analysis on free engines, with direct output to CDISC datasets, define.xml, reviewer's guides and a Module 5 package — for small labs and CROs that cannot justify the licenses of the big suites.
Apply for the pilotm5/datasets/phaxton-study/ ├ tabulations/sdtm/ │ ├ dm.xpt pc.xpt ex.xpt ✓ md5 │ ├ define.xml ✓ XSD │ └ csdrg.pdf ✓ fonts embedded ├ analysis/adam/datasets/ │ ├ pp.xpt adrg.pdf ✓ md5 │ └ define.xml △ no ADaM rules in CORE — declared ├ stf-phaxton-study.xml ✓ TCG file-tags └ index.xml ✓ backbone · ICH DTD CDISC CORE: 123/180 rules executed · 57 not executed, listed
Three steps, one wizard
A 7-screen web wizard walks the whole flow. Nothing to install at the lab.
Upload the CSV
Concentration–time data from your study. The input check looks at columns, units and gaps, and explains in plain words whatever does not match.
Analyze
NCA and bioequivalence on free engines. The NCA engine is checked parameter by parameter against the reference output shipped with Phoenix WinNonlin (12 profiles, 25 parameters).
Package
XPT datasets, define.xml, reviewer's guides and a Module 5 package with backbone, checksums and STF — checked by our own form checker before it is delivered.
What is included, and in what state
Every piece declares its state. "Validated" only when two independent arbiters agree; otherwise we say what it is and why.
| XPT datasets two arbiters | DM, PC, PP, EX and TS written through xportr (R), with two independent readers from different code families checking every file. |
|---|---|
| define.xml 2.x | Generated from the same source as the datasets — the same fact is never written twice. |
| CDISC CORE integrated | CDISC's official rules engine runs on the package, and the report always says how many rules it executed and how many it did not, with the reason. Never a "conformant" without saying against how much. |
| ADRG / SDRG | Reviewer's guides from templates, with font embedding checked as each PDF is written. |
| M5 package, eCTD v3.2.2 | Backbone checked against the official ICH DTD, MD5 checksums, STF with the file-tags from FDA's Technical Conformance Guide. Single sequence; multi-sequence lifecycle is out of scope and declared as such. |
| eCTD form checker PARTIAL | Checks form (structure, backbone, checksums with a second arbiter, references, PDF, names, STF) and repairs only on a copy, logging every change. PARTIAL: not yet contrasted against an agency validator. |
"What lacks evidence is declared as debt, not marked green."
See the circuit
A recording of about a minute and a half of the wizard, from a concentration–time CSV to the draft package. It was recorded on a local run with synthetic data; the interface is in Spanish, the captions and the narration are in English, and the long waits are sped up (the real run takes close to four minutes). The narration is a synthetic voice. PHAXTON is not hosted online: this is a recording, not a live demo.
Duration: 1:20 · WebM, 1437 KiB · English narration, synthetic voice (Piper, en_US-norman-medium, a male voice trained on public-domain recordings) · Download the video
Text of the narration
- This is the PHAXTON wizard, recorded on a local run with synthetic data. The interface is in Spanish. Step one: upload a concentration time CSV.
- Map the columns, and declare the units: route, concentration, time and dose.
- The input check reports what it found, in plain language. Here the status is yellow.
- Step two: choose the analysis. Here, non-compartmental analysis.
- Step three: configure the engine, and run it.
- Step four: run the analysis, and read the results.
- Step five: the regulatory report. Generating the package files takes a while, so this wait is sped up ten times.
- The report is delivered as a package. It is a draft.
- Step six: run the product's own checks. These are not CDISC CORE.
- On this synthetic data they report 1 error, 1 warning, and 4 of 6 rules passed.
- Step seven: the final package. Each piece is shown as partial or missing, with the notice that this is a preparatory tool. CDISC CORE was not run in this recording.
- Downloading the package takes about a minute, again sped up ten times.
- That is the circuit. The data are synthetic and the run was local. The output is a draft package, not a submission.
Example package
Download an example package built from synthetic data: a draft eCTD-ready Module 5 (SDTM and ADaM XPT datasets, define.xml, SDRG/ADRG reviewer's guides, listings and check reports), with the state of each piece stated as it is. Every piece of the package is marked PARTIAL or MISSING: CDISC CORE runs on the SDTM and reports how many rules it ran and how many it did not, but it has no rules for ADaM, and ADNCA does not fully match the ADaM IG for NCA. This is not a submission.
Download the example package (.zip)
Size: 538,454 bytes (525.8 KiB)
SHA-256: 3607dba040be43e6cc184dd8657c6cbb6f8613e24f03a3658c29096d1dd2c016
Free eCTD form checker
Phaxton's checking layer is released as a free, open-source tool (GPL-3.0): it checks the form of an eCTD v3.2.2 package — backbone, checksums, references, PDFs, study tagging file — and repairs on a copy, never on the original. Status: PARTIAL, not yet contrasted against an agency validator.
What Phaxton is not
- It does not submit to any agency: it prepares the package; submission is done by the applicant or their gateway provider.
- It does not guarantee acceptance of any submission, and it is not endorsed by FDA, EMA or any other agency.
- It does not assess scientific content: it checks form and reports diagnostics with cited rules.
- It does not replace review by a qualified professional — it requires it.
Pilot with small labs
We are looking for a small number of labs and CROs to pilot the full flow on their own studies. Write to us and we will walk you through it.
cperapa@fractalpk.es