Pass an Evidence Item Only When Normal Operation Produces It

Evidence by Operation Checklist

Companion to Medical Device Connectivity · Updated 2026-09-29

Use this when you define what the device records, and again whenever a customer, auditor, or investigator asks for evidence that someone had to assemble by hand. Product, quality, regulatory, and security engineers own it; “Make Devices Produce Their Own Compliance Evidence” works it on the sterilizer.

Each item passes only when normal operation produces it, in machine-readable form, and the institution can export it without asking the manufacturer. List the records customers, surveyors, and investigators ask for. Name the device event that produces each one. Record who uses each item and whether it passes.

The Evidence by Operation Checklist: evidence that should fall out of normal operation.

EvidencePass whenWho uses itStatus
Cycle or treatment recordsEvery cycle or treatment is recorded with its identities, parameters, results, and outcome, with no transcription by an operator, and joins automatically to related records held in other systems (such as load contents and indicator results for a sterilizer, or the order and specimen for an analyzer).
Software inventoryThe device reports its software versions and components on request and after every change.
Self-test resultsEvery self-test is recorded with its time and result.
CalibrationCalibration state and history travel with the data they affect.
Update historyEvery installation, failure, and rollback is recorded with its time and version.
Security eventsSecurity events are logged in a format the institution’s security monitoring tools can consume.
UtilizationUse and idle time are recorded per unit.
Failure recordsEvery fault is recorded with its error code and the device state at the time.
Configuration historyEvery configuration change is recorded with who made it and when.
Overrides and exceptionsEvery override is recorded with the operator, the time, and the reason given.

Put every failed row on the roadmap with a date. Until it passes, someone at the institution rebuilds that record by hand before each audit or investigation, and the manufacturer can’t use it to learn how the product behaves in the field or to support its own verification and regulatory work, which “Put AI to Work First on Verification, Validation, and Regulatory Evidence” covers.