This paper proposes an object oriented model to implement medical Knowledge Base Systems (KBS). The model has two levels. The external level (type level) facilitates the implementation and the interoperability of the system. The internal level describes the objects which are combined with the composition (union) operator. A fuzzy vectorial space defines the evolution of the clinical objects involved in the system. Thus, the composition of the clinical picture objects is tuned by fuzzy functions whose values evolve with time. The model provides new tools to compare the evolution of objects and to implement analogical reasoning.