Life Sciences / Pilot path
BioChangeGate · bounded pilotWhen scientific evidence changes, know what needs to change with it.
EcoKure can sit around a non-diagnostic research or evidence workflow to identify affected analyses, route uncertainty to a named reviewer and preserve a before/after record for inspection.
Demonstration scope: research and evidence operations only. Not a medical device, diagnostic, clinical decision-maker or regulatory certification.
One source update. One impact decision.
What the pilot demonstrates
- Which source or dataset revision changed
- Which downstream evidence is affected
- Which analyses can be preserved and which need replay
- Where evidence is incomplete and a human must decide
What stays out of scope
- Clinical diagnosis or treatment recommendations
- Claims of medical-device approval or legal compliance
- Automatic promotion into a production clinical workflow
- Replacing the scientific or quality owner
From readiness evaluation to evidence.
Choose one workflow, owner, source boundary and success/failure criteria.
Translate provenance, revision, confidence and review requirements into versioned controls.
Observe the workflow without acting on clinical or production decisions.
Introduce a controlled source revision and classify the downstream impact.
Replay only affected deterministic analyses and route exceptions to a named reviewer.
Produce a board-, quality- or audit-ready before/after record.
Customer owners decide whether the evidence supports the next stage.
A real signed chain, using non-customer data.
This page now runs a controlled two-revision fixture through the same shape of evidence chain a pilot would inspect. The signing key is demo-only; the cryptographic verification is real, but no production or customer claim is being made.
f31e8a7b2d3ea9b6…baseline9453e5a0cdfa3e7c…changed revision| Chain record | Result | Why it matters |
|---|---|---|
| 01 · | 3b102076f7b025f6cf… | Signed and linked to the previous record. |
| 02 · | 3aa091fd46d9b57bb9… | Signed and linked to the previous record. |
| 03 · | 0ef764d409f5b53248… | Signed and linked to the previous record. |
| 04 · | 0f3300d4600bddb45a… | Signed and linked to the previous record. |
| 05 · | bd1a3e909513f71603… | Signed and linked to the previous record. |
Evidence that can be inspected later.
| Record | Example | Why it matters |
|---|---|---|
| Source identity | Dataset/version/checksum | Establishes exactly which evidence was used. |
| Dependency impact | Affected / preserved / unknown | Prevents a broad re-run from hiding the actual change radius. |
| Control evaluation | Provenance, confidence, review state | Makes the operating rule explicit and versioned. |
| Replay record | Scope, result, environment | Shows what was rerun and what was deliberately preserved. |
| Reviewer action | Identity, reason, disposition | Separates machine classification from qualified human judgement. |
Start with one evidence change.
Bring one bounded, non-diagnostic workflow and we will define the readiness evaluation before discussing production.
Scope the pilotWhat the worked fixture returns.
This is the shape of the review a customer would receive. The values are from a non-customer demonstration fixture; customer owners would set the acceptance criteria and make the go/no-go decision.
Impact set produced
Affected, preserved and unknown records are separated before replay.
Uncertainty is held
Unknown provenance is routed to qualified human review instead of guessed through.
EvidencePack verifies
Signed records link the source revisions, impact decision and replay result.
Decision status: DEMONSTRATION COMPLETE · CUSTOMER PILOT PENDING
Next step: agree one customer workflow, owner, source boundary and written success criteria.