Many AI products have been cleared through Regulation and may also have been evaluated prior to deployment at the site - see how we help with Evaluation here. But once in use, there is a further need to monitor their performance continuously over time.

Products in use may undergo regular developments and upgrades, raising questions on which version should be used. But even without any change to the AI software itself, its results may change over time as data inputs to the product change. This is known as data drift.

For hospitals.

A hospital using AI needs to know that the product keeps performing well and safely for its patients over time, and be alerted to any unexpected changes and why those might have occurred.

For AI developers.

AI vendors need to know how their product is performing to be able to improve their offering and to satisfy their post-market surveillance and clinical follow-up obligations.