Algorithmovigilance
Also known as: Algorithm Surveillance
The ongoing surveillance of deployed clinical algorithms for performance drift, unintended effects, and harm, modeled on the pharmacovigilance practice used for medicines.
Algorithmovigilance names the discipline of watching a clinical algorithm after it is in use, on the premise that a model validated once is not validated forever. Patient populations change, upstream data changes, clinical practice changes, and the model itself may be updated by its vendor.
The term borrows deliberately from pharmacovigilance, which established that post-market surveillance of a product is a distinct obligation from pre-market approval.
This is a young practice and organizations implement it inconsistently. What is consistent is the direction: health systems are moving from "did this pass evaluation" toward "how will we know if it stops working."
Vendors who can supply monitoring data, drift alerts, and revalidation support are answering a question buyers are increasingly required to ask.