External validation
Testing a model on data from a different hospital, region, or population than it was trained on. It is the real test of whether a model generalizes; performance almost always drops compared to internal testing.
A department reviewing a fracture detection tool finds that every performance figure in the supplier's evidence pack comes from a held-out slice of the developer's own two hospitals. Those test patients shared equipment, imaging protocols and labeling conventions with the training data, so the figures describe something close to a best case. Ask whether the tool has been tested at any site the developer did not train on, how far performance fell there, and whether that site resembles your own more closely than the original two.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →