TERMINUS reads the logger data, identifies the excursion, runs the Arrhenius assessment against your own stability data and returns a QA/QP recommendation — with the audit trail already attached.
Every excursion review answers the same four questions. The work is not the thinking — it is the assembling. That is the part TERMINUS takes.
Logger files and ambient sensor data are read as they come, without retyping into a sheet.
Time above and below the labelled range, with the exposure window isolated to the minute.
Arrhenius assessment with activation energy from your own ICH Q1A stability studies, then a post-exposure shelf-life simulation — not a rule of thumb. The method, worked through →
A recommendation with the reasoning and the audit trail attached. The decision stays yours; the assembling does not. Who decides, and on what evidence →
A 2–8 °C shipment spends 1 h 30 min above 8 °C, peaking at 9.3 °C. The question QA has to answer is not whether the limit was crossed — the logger says that. It is what the exposure did to the remaining stability.
Activation energy comes from the stability studies for that product, not from a generic constant. The post-exposure simulation returns 10 days of shelf life lost, which moves the expiry on the label from 12 Mar 2028 to 02 Mar 2028.
That last line is the one an inspector reads. It is also the line a spreadsheet almost never produces, because the spreadsheet stops at “excursion: yes”.
Two adjacent methods worth knowing before you rely on either: how MKT is calculated, with a worked example and Arrhenius, ASAP and AccelStab compared.
Illustrative example built on simulated logger data. Your own figures depend on your product, your stability data and your lane.
| Step | With TERMINUS |
|---|---|
| Logger data pulled and pasted into a sheet | Read directly |
| Excursion window worked out by eye | Isolated from the raw series |
| Ea looked up, sometimes assumed | Sourced from your ICH Q1A studies |
| Impact judged qualitatively | Post-exposure shelf life simulated |
| Justification written from memory | Audit trail generated with the decision |
TERMINUS does not replace the systems you already run. It does the assessment that sits between the logger and the QA decision — the part that today lives in a spreadsheet, an email thread and one person's memory.
The first three questions a quality or validation team asks about a tool like this, answered before you have to ask them.
It produces an advisory recommendation — release, conditional or hold — with the reasoning and the evidence attached. The formal disposition is made and signed by your Authorised Quality Reviewer under e-signature, and recorded in your QMS. That gate does not move.
Configurable software validated through a risk-based lifecycle: intended use, configuration, requirements traceability, IQ/OQ/PQ, data-integrity and audit-trail verification. We supply the product and supplier evidence; you own the validated state for your intended use. Why Category 4 and not 5 →
Language models are used only to read what is unstructured — a stability PDF, a logger layout nobody has seen before — and every extraction is confirmed before it is used. The assessment itself runs on deterministic, version-pinned rules and Arrhenius kinetics.
The number in the headline of this page, with the method and the anonymised pilot data behind it. Written for people who will check the working.
A 30-minute working session on a real case from your lane — your data, your stability studies, your call at the end.
