At a minimum, the characteristics of local data should be compared with those of the test data (a typical example might be where a model has been tested only on one manufacturer’s MRI scanner, but will be used on a scanner made by a different manufacturer). Where these are similar, ...