Fuzzy hypothesis management for disorder diagnosis

We propose the use of a knowledge based framework for diagnosis in which the knowledge base consists of particular instances of general disorder models. We study how to select which manifestation (symptom, malfunction) to query in order to reduce a set of competing diagnosis hypotheses (disorders), none of them completely satisfying, considering only the observed manifestations. We propose to use similarity relations to guide us on the task of selecting which manifestations to further investigate in order to confirm or eliminate a hypothesis, as well as information about the order in which the competing hypotheses should be probed first.