Abstract

The nosological diagnosis is a particular kind of a-theoretical diagnostic process, which aims to recognize the pathological condition of a patient on the basis of symptoms and clinical signs, without using a pathophysiological or etiological explanation. In this work, we consider nosological diagnosis as a form of clinical reasoning composed of an inductive categorization phase followed by a phase of analogical classification based on similarity; then we show how an automatic classification system, the PELC – Prototype Exemplar Learning Classifier, can realize a cognitive-computational model of this type of diagnosis. The proposed model is based in part on prototype resemblance theory of disease proposed by Sadegh-Zadeh, and in part extends it including also exemplar theory developed in cognitive science. Moreover this automatic classification system can also be used, thanks to its cognitive plausibility, to extract nosological knowledge (e.g. inferring syndromes and atypical clinical cases) from clinical datasets.