Modelling of Longitudinal Information Systems with Graph Grammars

Longitudinal information systems (LIS) manage and evolve data over extensive periods of time. Examples are "womb to tomb" electronic health records. How can we design such systems such that they are future-proof, i.e., evolvable in step with changing requirements? One approach that has been advocated is the "two-level modelling" approach, separating information and knowledge in terms of a small reference model and a larger archetype model. A textual archetype definition language has been proposed to define the mapping between these two models. In this paper, we explore an alternative way to define this mapping using triple graph grammars. The graph grammar based approach has several advantages over the textual approach, including better modularity and tool support.