The analysis of the validation survey was approached from three angles: reviewers, validation questions, and test items.
External validation is usually approached by comparing questionnaire data with data measured by other methods such as visual inspection (8 – 10 ), photographs (11 ), selection of clinical signs (2, 4 ), laboratory test results (12 ), or other (4, 13 ).
Robustness Validation is an approach to a failure-mechanism-driven qualification that was primarily initiated by the automotive industry but has a much wider range of applicability.
QTL validation is another approach to confirm the effect of QTL across different genetic backgrounds.
Manual validation is the gold standard approach to confirm accuracy of database identifications, but is extremely time-intensive.
Further validation is required before this approach can be recommended for clinical practice.
Using this approach, validation is a process that involves accumulating evidence to provide a strong scientific basis for proposed score interpretations.
This indirect approach of validation is also known as 'pattern oriented modelling' [30, 31], 'indirect' [29] or 'secondary predictions' [28].
The practical and ethical difficulties of validation by clinical trial may be insurmountable, and an alternative approach to validation is proposed which does not raise these difficulties.
The level of data validation is minimal and sufficient for testing; a comprehensive data validation approach has been outlined for implementation prior to production release.
A separate analysis is presented in online Appendix B, in which the bootstrap resampling approach to model validation was compared with an approach that used training and validation data subsets.
