All apps

Clinical QMS · Clinical QA / Quality managers

CuRE Credo

Clinical-GCP QMS — deviations, CAPA, training, and inspection-ready quality evidence.

What it does

Clinical-GCP QMS — deviations, CAPA, training, and inspection-ready quality evidence. QMS evidence is linked to the same governed study and operational record as the rest of CuRE, not parked in a separate quality silo.

Key capabilities

  • Site-scoped deviation register with category codebook
  • CAPA lifecycle management
  • Training assignment evidence
  • Inspection-ready quality traceability
  • Study-data-grounded root-cause analysis
  • Automated deviation intake from Capture
  • CAPA recurrence-linkage
  • Management-review cockpit
  • Quality trending & analytics

What sets it apart

  • QMS evidence is linked to the same governed study and operational record as the rest of CuRE, not parked in a separate quality silo.
  • Root-cause analysis queries the actual OMOP subject/visit/site data — an RCA no document-only QMS (MasterControl, ETQ) can run.
  • Supplier/CRO oversight, validation-package custody, change control, LMS/courseware, and HA-finding loops complete the QMS.
  • CAPA and training workflows can inherit signals from Caliber, Canary, Capture, and Conduct instead of starting from manual re-entry.
  • Built on the platform's audit, e-signature, and artifact primitives, so quality records share one validation story.
CuRE Credo · deviation → CAPA routing
Live demo — synthetic data, runs in your browser

Triage a deviation and watch it route to a CAPA

A GCP quality system doesn’t treat every deviation the same. Set the three facts a QA reviewer weighs — subject-safety impact, data-integrity impact, and whether the problem is systemic — and watch Credo classify the severity, then route the disposition to a tracked CAPA, a deviation-only record, or an info-only acknowledgement. When it opens a CAPA, step it through its closed-loop lifecycle and bind its effectiveness check to a live monitoring metric so data, not an assertion, decides whether the fix held.

A synthetic IND — ravucizumab (VELT-042). Pick a quality event, then set the facts to watch it classify + route.

DV-2041IMP stored 6 h above labeled range at Site 004

A temperature excursion logged the investigational product at 12 °C (labeled 2–8 °C) for ~6 hours before the fridge alarm was actioned. Two dispensed kits are affected.

Subject-safety impact
Data-integrity impact

These are the facts a GCP QA reviewer weighs against the study’s deviation plan — the classification and the routing derive from them, not from a stored answer.

Classification & routing

Severity
Major

Materially affects a subject’s rights, safety, or well-being, or the study’s primary data, without being critical.

Routed to
CAPA

Open a CAPA · corrective

  • Major severity requires a tracked corrective action (CAPA).

The routing precedence is the product’s own — a CAPA linkage wins over a deviation-only record, which wins over an info-only acknowledgement (capaId ? "capa" : deviationId ? "deviation" : "info_only").

Deviation record · Part-11 transition path

Classifyclassified
Dispositiondispositioned
Closeclosed

open → classified → dispositioned → closed. The disposition and close are §11.10(g) e-signed and fail closed — a refused re-auth rolls the whole transition back. Each transition is sealed into the Part-11 hash chain.

CAPA · closed-loop lifecycle

corrective
open
in_progress
effectiveness_check
closed
Root-cause analysis

Iteratively ask “why?” down a causal chain to the root cause.

Effectiveness check · bound metric
pass
Cold-chain excursions per month (Site 004)
Baseline
2
Threshold (≤)
0
Observed
0

The effectiveness check binds to a live monitoring metric (a Caliber KRI/QTL). When a fresh reading lands, Credo compares the observed value to the acceptance threshold — pass when observed ≤ 0 excursions/mo — and flips the verdict automatically. Data, not an assertion, decides whether the fix held. The QA owner still §11.10(g) e-signs the close.

This study’s deviation queue

DeviationSeverityRoutes to
DV-2041 IMP stored 6 h above labeled range at Site 004MajorCAPA · corrective
DV-2042 Primary-endpoint assessment window missed for 9 subjectsCriticalCAPA · preventive
DV-2043 Consent re-signature captured one day lateMinorDeviation
DV-2044 Eligibility lab drawn at wrong visit (recurring)CriticalCAPA · preventive
DV-2045 Unblinded pharmacist viewed randomization list prematurelyCriticalCAPA · corrective
DV-2046 CRF signed with a stale delegation-log entryMinorInfo-only

Every classification, routing decision, and effectiveness verdict is computed in your browser on synthetic facts — no backend, no real data. Click a row to load it. Notice the routing is never automatic: it is grounded in the deviation’s own safety, data-integrity, and recurrence facts — the same facts a GCP inspector expects the QMS to have weighed.

Why this is more than a toy

The routing decision, the transition guards, and the effectiveness check are a faithful, dependency-free port of Credo’s real methodology, from source files under apps/credo/src/server/: the deviations/router.ts transition machine (open → classified → dispositioned → closed, with the §11.10(g)-e-signed, fail-closed disposition and close), the quality-events/router.ts routing precedence (capaId ? "capa" : deviationId ? "deviation" : "info_only"), the capas/ lifecycle machine (open → in_progress → effectiveness_check → closed) with its corrective/preventive kind and 5-why / fishbone RCA discriminator, and capas/effectiveness.ts’s evaluateEffectivenessCriterion() — the “data-closes-the-CAPA” predicate — ported line-for-line. The severity grading is the GCP critical / major / minor criteria (safety, data-integrity, systemic scope) made legible; in-product that is a QA judgement, not an auto-score. Per ADR-PLT-044 all of this is Credo’s own event-write-bound operational state — a deterministic state machine + threshold check, not statistical methodology — so it lives in-app rather than in Calculate. Here it all runs entirely client-side on a synthetic IND — no backend, no real data.

See CuRE Credo in action

Every research ecosystem is unique. Let's discuss how CuRE can be configured for your needs.