Evidence chain
One ALT value, from a scanned PDF to a signed decision.
Six states, each released through a gate, each carrying where it came from. Click any step, or walk it in order.
PDF / CART001-0042 / ALT
Central-lab report
A scanned central-lab PDF. On page 2, in the liver panel, the ALT row is flagged H.
ARTIFACT → ARTIFACT_SEGMENT
| artifact_id | art-0042-0917 |
| artifact_type | central_lab_report |
| artifact_name | LAB-20260917-0042.pdf |
| source_system | central lab · site 03 |
| file_hash | 9f3c…e71a |
| segment_id | seg-0042-0917-p2-r3 |
| segment_type | table_row |
| locator_json | {"page":2,"table":"liver_panel","row":3} |
| raw_text | ALT 128 U/L H (7-40) |
Derived from
SOURCE_BATCH sb-2026-0917-03 · sFTP connectorVia
segmenter v2.3Received as-is
Learning loop
The signature that corrects the draft also guards the next change.
Every correction becomes a regression case. A rule or threshold cannot change until it passes them. Continue the same case.
Sign / POL-LFT-02 / causality.lft
A signature that changed the draft
Back at the end of the previous chain. The co-pilot drafted causality as 'possibly related'; the medical monitor changed it to 'probably related' and signed. The change and its reason are in the ledger.
DL-2026-0918-031
| subject | CART001-0042 |
| event | ALT increased · CTCAE G2 · Day 7 |
| approver | Medical Monitor · demo account |
| causality | Possibly related Probably related |
| rationale | Grade 2 ALT rise; alternative causes not excluded Grade 2 ALT rise 7 days post-infusion with concurrent CRS G1; pattern consistent with CRS-associated transaminitis |
Outcome
Signed. The change is flagged as a correction.