A framework for chronic kidney disease diagnosis based on case based reasoning

Chronic kidney diseases are very critical. Case-based reasoning (CBR) is a reasoning technique suitable for problems that depend on experiences. The first step in building CBR system is preparing a comprehensive case base from patients' electronic health records (EHRs). EHR data need quality improvement steps, such as normalization, feature selection, feature weighting, and outlier detection. In the medical field, the representation of resulting case base using formalized concepts and terminologies is highly needed. There are many structures for representing case bases, but the most powerful method for representation is using ontologies. The manuscript proposes a methodology for diagnosing the chronic kidney disease based on an ontology reasoning mechanism. In this paper, we first prepare the chronic kidney dataset of 400 real cases with 25 features by utilizing a set of data mining algorithms. Next, we construct an ontology structure to represent this case base in the W3C web ontology language (OWL) ontology format and populate this ontology with the individual cases. The fuzzy rough set algorithm achieved the highest accuracy for selecting the most suitable feature set. The resulting OWL ontology is based on disease ontology (DO) semantics, which is the most common and standardized ontology in the medical field.