Analysis of clinical documents to enable semantic interoperability

While there is a wealth of information available for each patient in an Electronic Health Record (EHR), information is not optimally organized for efficient use in patient care and there is still a large gap to achieve semantic and process interoperability. Current researches focus either on fully structured or non-structured clinical documents. A pre-analysis of 1000 real-world clinical documents showed, that most clinical documents are provided in a semi-structured way as level 2 HL7 Clinical Document Architecture (CDA) documents. Thus, an analysis framework is presented which also uses this semi-structure in combination with underlying models and metadata provided by the exchanging infrastructures, allowing summarization, comparison and integration of clinical information. The approach was tested using an IHE compliant EHR system. The analysis framework offers good response times and high accuracy levels. However, there is room for improvement considering processing of non-structured text and non-CDA documents.