Recommendation	Rationale
Knowledge engineering techniques may be considered as an adjunct to the creation of custom groupings and SMQs designed for the selection and extraction of case reports in pharmacovigilance databases	The PROTECT studies show it is possible to propose relevant novel groupings when no predefined grouping is available in MedDRA® for a given safety topic (e.g. anaphylactic shock or upper gastric hemorrhages) [16, 17]
Additional research would be necessary to validate if novel groupings generated by knowledge engineering techniques can help in the design of appropriate groupings of MedDRA® PTs for use in signal detection or evaluation	Given the current state of research, the clinical accuracy of groupings generated by knowledge engineering is such that manual clinical review is still required and this still needs to be validated against existing signal detection methods [16, 17, 19]
Consideration should be given to piloting the use of knowledge engineering in developing groupings in other ontologies for application to other vocabularies and their possible linkage	Given that it has been shown to be possible to generate relevant novel groupings in MedDRA®, it is reasonable to expect that it would also be possible in other ontologies, e.g. ICD10, SNOMED