Figure 3

Linked clinical data cube architecture aligned with the RDF data cube. Depicts the architecture of the LCDC. The Main cube (depicted as a red cube) is split into modular Specialised cubes (depicted as a blue cube) and linked using the void:subset property. The Main cube is organised into time-series, cross-section and theme slices using the qb:slice property. The slices are then divided into Observations using the qb:observation property. The qb:dataset property is used to link the observations back to the cube. The Specialised cubes are organised similarly to the Main cube with the exception of the theme slices. The dotted lines show how the slices from all cubes interlink to the study observations through the use of ObservationGroups and the qb:observationGroup property. The mainObservation property manages the link between the ObservationGroups and the Observations while the specialisedObservation property handles the link between the ObservationGroups in the main cube and the corresponding Observations in the specialised cubes.