Decision support systems from the standpoint of knowledge representation.

Relationships between decision-support systems and knowledge representation are examined from three different points of view: the characteristics of medical decisions that might influence the selection of appropriate knowledge representations,--the extent to which different knowledge representations can support efficient medical decisions and,--the validation of knowledge hypotheses through the practice of decision support systems. A three-level model of knowledge representation is proposed that includes a contextual, a conceptual and a computational level. Taking into consideration the context that leads to the selection of a given representation raises the issue of multiexpertise and multirepresentation modeling. Implementation of decision support systems as sets of cooperative agents and integration in the health information systems are considered.