Automatic transcription of electronic medical records to case structures for use in medical case base reasoning systems

摘要

Case-Based Reasoning techniques provide 'analogy-based' solutions to clinical problems by manipulating knowledge derived from similar previously experienced situations, called Cases. Efficacious application of CBR based systems in healthcare demands a continuous supply of up-to-date and correct (clinical) cases (in an electronic medium) from medical experts, which is perceived as a drawback since medical experts are usually hard pressed for time and resources to handle such tasks. To address this constraint, we present an automatic and functionally robust strategy to proactively transform generic Electronic Patient Records (EPR) to Operable CBR- oriented Cases (OCC) that &e compliant to specialised CBR-based medical systems. EPR-to-OCC mapping is based on match-identification-between EMR document-objects and case meta-structures-at the following levels: (1) explicit attribute-value object descriptions, (2) equivalent-terminology descriptions, and (3) equivalent-concept descriptions. The above-described strategy is implemented via a Java-based computer system that incorporates computational mechanisms based on XML parse-trees, Unified Medical Language Source (UMLS) meta-thesauri and medical knowledge ontologies. In conclusion, the automatically transcribed OCC can be seamlessly incorporated within Intelligent CBR-based Medical Diagnostic Systems.