Glossary
Model Drift
Also called: performance drift, concept drift
Gradual degradation of a deployed model's performance over time as clinical practice, populations, or equipment change around it.
Model drift is dataset shift arriving slowly. A scanner is replaced, a protocol is updated, a referral pattern changes, coding practice shifts. The model does not change, its environment does, and its performance moves.
Where This Gets Misread
Drift is only detectable with ongoing monitoring against outcomes, which requires infrastructure most health systems have not built. A model deployed and left alone is not stable, it is unmonitored, and those are not the same thing.