Ontology-driven approach to medical data fusion

A large volume of operational and statistical data has been stored within medical information system databases. The stored data represents a loosely-coupled and semi-structured clump. To increase information support of medical procedures it is necessary to implement data linking processes into the medical information systems. The aforementioned processes should be able to collect data from diverse sources and produce a linked view that can be used for medical evaluation, planning, etc. To solve this problem, an approach based on JDL (Joint Directors of Laboratories) data fusion model and ontologies is presented. JDL model defines functional levels of data fusion and their interconnections. Ontologies are used to set links between data and knowledge. To specify and evaluate the proposed approach, a simplified scenario is constructed. The scenario was prepared for the Federal Almazov North-West Medical Research Center. The proposed scenario includes the following actions: extraction of a list of drugs and diagnosis from semi-structured data; annotation of extracted data using author's ontology and third-party ontologies; revealing interacting drugs that are prescribed to a patient; data representation. The scenario shows that ontology-driven data fusion approach has a great potential for linking data and knowledge within medical information systems. It helps to detail each level of data fusion, reveal problems that need further research and describe use cases for Semantic Web instruments.