Drift monitor
Re-tests on a schedule and flags real changes.
Checked against clinicians
Alert precision against clinician-confirmed regressions.
Where a person decides
Alerts to owner; no auto-rollback of clinical systems without approval.
Works on
All.
Delivered inside a Quality Audit today. Reviewed .
The problem it solves
Vendors change models, prompts and speech engines. Quality can slip without anyone noticing until clinicians complain.
How it works
Run a fixed suite
A frozen case set runs on a schedule and after known changes.
Repeat each case
Repeats separate real change from run-to-run noise.
Alert on real shifts
Alerts fire when a change exceeds normal variation, with the cases that moved.
An example, illustrative
Everything looks the same in daily use.
Alert: dose-error rate up from 2% to 6% after Tuesday's model update.
A baseline case set and access to the product.
It detects change on your suite. It can't see problems the suite doesn't cover.
What it helps measure
Quick answers
What does the drift monitor do?
Scheduled regression runs with repeated trials after vendor, model, EHR or ASR updates.
How is the drift monitor checked against clinicians?
Alert precision against clinician-confirmed regressions.
How often does it run?
Monthly by default, and immediately after announced changes.
Can it find the cause?
It can hand the moved cases to the replay agent for diagnosis.
What can the drift monitor not do?
Alerts to owner; no auto-rollback of clinical systems without approval. No agent makes a clinical decision.