Anemia: An artificial intelligence project

This paper presents a survey of the history of ANEMIA, an artificial intelligence project for building a computer based consultation in the field of anemia. The knowledge base was represented by production rules, and EXPERT, an artificial intelligence programming scheme developed at Rutgers University (USA), was used as software framework. A first model, based on the selection of a pivotal finding (MCV) as the general criterion for arriving at the diagnostic categories proved early to be too rigid. In a second model the disease categories were profiled in terms of intermediate hypotheses and organized into a taxonomic hierarchy. At present, the knowledge base of ANEMIA is derived from 220 patients records.