Optimising cohort data in Europe

mechanism is needed, it is necessary to identify the context where these resources for common datasets emerge. Minimal datasets do not emerge out of anywhere: it is only possible to generate themonce there is sufficient metadata about what other researchers are using (in terms of common data elements). Thus, the knowledge integration mechanism should be relatively straightforward and operate through existing procedures and standardised information (i.e. rules and directives) that will regulate the generation of minimum datasets. The appropriate knowledge integration mechanism is thus to take into account how others use metadata and generate common codes for data structure and make an informed selection for a minimal dataset. However, it is also possible on some occasions to start from a common minimum dataset. The JRC (Joint Research Center for rare disease in EU) defined common data elements that are now referenced by all the 24 European reference networks for data comparison. In summary, metadata standards suffer from a lack of connection between local, specialised resources and global versatile resources. The same lack of connection is also apparent in current harmonisation levels. Namely, in order to determine the level of harmonisation needed, it is necessary to first determine the kinds of resources concerned. On the one hand, the harmonisation level depends on the specific purposes for which the research is being carried out. On the other hand, it is crucial to understand what is to be harmonised in the first place. That is, harmonising data is far more difficult than harmonising metadata. This is because data harmonisation does not directly concern data per se; hence, it is impossible to fix the data items themselves. What is actually harmonised are data metrics and organisation, where suffixes and prefixes link data points to definitions. Data points are thus needed to determine the harmonisation level, but data points are highly specialised resources, which limits their application. That is, each study has its own data points with specific data items. This may lead to resource immobility: the organisation of data and data points cannot be transferred to other cohort research projects. There are no established standards for data harmonisation levels, especially when innovative questions are concerned. Thus, the question is, which knowledge capability can best confront such specialised and potentially immobile resource so that the harmonisation level can still be determined. We suggest that integrative capabilities of knowledge aggregation and knowledge transferability are needed. Such integrative capabilities will not fundamentally change data points, but will allow to categorise, structure and define data points without changing them. In summary, Pillar I converts specialised resources into versatile ones through integrative capabilities (e.g. aggregation). An example of such a process is when data from multiple