Ontology-Driven Health Informatics System for Intelligent Clinical Knowledge Representation and Reasoning

Abstract

The increasing amount and diversity of clinical data present serious problems to efficient knowledge representation and smart decision making in contemporary healthcare systems. Traditional data-based methods do not typically provide the means to express semantic relationships and domain knowledge and are more restrictive in their ability to be interpreted and used by clinicians. This paper provides an ontology-based health informatics infrastructure of intelligent clinical knowledge representation and reasoning. The suggested framework makes use of the semantic web technologies with the help of Web Ontology Language (OWL) and Resource Description Framework (RDF) to model a structured clinical knowledge in the form of well-defined entities and relations. The ontology is combined with a rule-based reasoner based on the Semantic Web Rule Language (SWRL) to support automated inference. The system is created with prototeg and the Pellet reasoner to verify and create logic. It is evaluated on MIMIC-III clinical dataset in its subset including the patient symptoms, diagnoses and the treatment records. Experimental evidence has shown that the proposed system has a reasoning accuracy of 91.2 with a better query response time and a better interpretability level than traditional methods of rule-based reasoning. The results of the study reveal the usefulness of ontology-based frameworks in facilitating scalable, transparent and knowledge-based clinical decision support systems.