A Decision Support System for Diabetes Mellitus Management

Abstract

DM is considered a dangerous chronic disease. Diagnosis is the first step in its management. CDSS for DM improves its detection and decreases the opprotunihty for complications. However, its diagnosis is a theory-less problem. CBR is a problem-solving paradigm that uses past experiences to solve new problems. Integratrion of CBR and formal ontologies enhances the intelligence of this paradigm. Utilizing patients' EHRs for building case-based knowledge solves the problem of knowledge acquisition bottleneck but preparation steps are required. Moreover, using standard medical ontologies such asn SNOMED-CT enhances the interoperatbility and integration of CDSS with the healthcare system. If ontology-based CBR systems utilize vague or imprecise knowledge, the semantic effectiveness is further improved. This paper proposes an advnaced and complete fuzzy-ontologist-based CBR framework that manages and utilizes imprescise knowledge. We implement the most critical steps in CBR (case representation and retieval). The implemented framework has been tested on the diabetes diagnosis problem using a case base of 60 cases from the EHR on Mansoura University Hospitals, Egype. The proposed system has an accuracy of 97.67%.