ONTOLOGY ENABLED CASE BASED REASONING FOR NEONATAL DISEASE DIAGNOSIS

Abstract

iseases that attack newborn babies within the first 28 days of their lives are referred to as neonatal diseases. It is still a major health issue today because today’s medical diagnosis depends on an understanding of human intelligence, which takes time and creates inconsistency in decision-making. To solve the issue, integrating expert knowledge with technology is required. Currently, knowledge-based systems have become a possible choice to support experts in decision-making, regardless of the diagnosis of various diseases. Both ontology and case based reasoning are types of knowledge based reasoning techniques used in this study to model knowledge and design prototypes, respectively. The goal of this study is to develop an ontology- enabled case based reasoning system that integrates ontology and case-based reasoning for the diagnosis of newborn diseases. To provide ontology-enabled case based reasoning, ontology was built independently and mapped to case based reasoning by using the Description logic extension, Onto Bridge, and Jena connector. A design-science research method is followed to structure the study. The knowledge needed for neonatal disease diagnosis was captured from the neonate dataset, which was collected from Haramaya University Hiwot Fana Specialized University Hospital. The collected dataset was preprocessed for duplication, cleaning, and transformation using Weka and Excel. Ontology is constructed using the protégé tool and owl language, while case based reasoning is designed using the Jcolibri framework. The protégé takes the captured data and represents knowledge in an organized and structured manner. Pure ontology is used to create case structure, and the ontology has been used in the case based reasoning application as a case base. The prototype scored an average of 88% accuracy, with precision and recall of 87% and 87%, respectively. Likewise, the user acceptance testing scored 89%, which shows the acceptability of the prototype by experts. This shows the prototype has registered a positive outcome and was well accepted by the experts to develop an applicable system to solve the problem. However, integrating DM results with the ontology to capture hidden knowledge and automatically update the case base needs further enhancement in the future