Granular computing in medical informatics

The intent of this study is to establish a comprehensive design environment of granular computing with emphasis on the analysis of complex decision-making process. Fuzzy modeling has revealed a useful tool for describing the various abstraction levels of the classification task. In particular, the different levels of data abstraction in the process of diagnostic classification in medical informatics have been considered. In the study, we provide taxonomy of granular models by distinguishing between descriptive and predictive models. Three representative examples of information granulation have been considered, that is, self-organizing maps, radial basis functions, and linguistic models. A certain complex problem of diagnostic classification (ECG classification in a database) has been considered as a case study. The obtained results demonstrate the potentiality and the powerful expression using the proposed methods.