Transformation of Archetype-based EHR into Semantic Model

Abstract

One of the main reasons for the adoption of Electronic Health Record (EHR) is to provide clinical decision support capability. The design of a typical Clinical Decision Support System (CDSS) is composed of three main models: information model, concept model, and inference model, and also of interfaces between these models. Among the models, information model gives structure to clinical data of EHR for interoperability. Initially non-standard and application specific information models were used for EHR, and later the emergence of health standards such as HL7, openEHR and CCR, introduced standard information models. The integration of so many information models, including standard and non-standard models, impeded the development of generic CDSS‟s. Therefore, the research calls for a common model that aligns various information models to provide a standard base access to decision support system. The unifying information model should be capable of capturing syntactic and semantic knowledge so as to be used in CDSS inference. We have proposed semantic model for the integrated view of EHR and associated metadata, and suggested model transformation technique to unify heterogeneous information models. In this work, the openEHR extract schema, ISO EN13606 extract schema, and HL7 CCD schema have been unified in ontologies using XSLT transformation. The syntactic correctness and completeness of the transformation has been verified with Altova SemanticWorks, while the logical consistency of the generated ontology has been validated with Pellet reasoner. In comparison to work by Isabel Román, we contributed an automatic approach for ontology generation from openEHR EHR extract schema, and also provided ontology generation for ISO EN13606 and HL7 CCD schema. The resultant ontologies not only provide syntactic but also semantics of the clinical data, and can be seamlessly integrated with CDSS for computer reasoning.