Temporal Validation.
In temporal validation, the model is typically validated in more recent
individuals. The purpose of such validation is to make sure that the
model maintains its accuracy when it is tested in cohorts in different
time periods. A model developed way back (say 20 years ago) may not work
in current patients (e.g., change in risk factor distribution and
availability of large dataset on many risk factors). Temporal validation
can be easily achieved just splitting the data into two parts. Develop
the model in one part that contains early treated patients and validate
the model to assess its performance in another part that contains
patients that are more recent2. Temporal validation
can also be achieved through the prospective application of the
developed model in specifically collected cohort2. For
example, a model can be developed in a group of patients between 2005
and 2010 and the same model can be validated in a different group of
patients from the same cohort between 2012 and 2015.