Different models of Decision Support Systems (DSS) are used in medicine to help physicians in disease diagnosis, prognosis evaluation and therapy prescription. The DSS models rely on mathematical or computer theories. Each of them offers advantages and drawbacks. We propose a DSS system architecture which tries to integrate the different kinds of decision models and to use them to deal with the successive clinical decision steps. Besides, the decision process includes patient data, experts' knowledge, statistical and epidemiological data, experience. The Case Based Reasoning (CBR) approach is used to store and retrieve the previous clinical cases. The object case components are clustered, indexed and stored in an object data base. To sum up, we propose a framework for clinical decision making and experience storing based on the main DSS models. Each step of the decision process is supervised by a finite state automaton which triggers the appropriate module and knowledge or data sources. At last, we illustrate our approach through an example : the epilepsy diagnosis and therapy.