Ontology based disease diagnosis system with probabilistic inference

Abstract

Disease diagnosis system integrates fragmented and constantly evolving information about diseases from various sources, such as the web, and builds up an ontology model and inference system, which could be used by doctors and researchers in the medical domain. It can also be used to make the information about rare diseases accessible for experts world over, which may not, otherwise, be easily available. This paper presents an ontology-based system that creates a knowledge graph of the extracted information to make it machine understandable. Instead of using deterministic logic, it generates a Bayesian network out of it that can be utilized to infer the most probable diseases given the patient's symptoms, age, gender, lifestyle and personal and parental medical history as input. It further recommends relevant diagnostic tests and pathology labs where the tests could be undertaken. It has a user-friendly interface through which it can take SPARQL query as an input and show the result to the user.