Seiteninterne Suche

Forschung

OntoImportSuite

 

The „OntoImportSuite“ is a set of proof-of-concept tools to extract and process heterogeneous data from clinical system, like the electronic medical record. This data can be mapped and transformed to match a set of common data elements, which will then be exported into a target system (i2b2) for research purposes. The system uses OWL ontologies to store information about the source and target systems and the mapping and transformation knowledge. The OWL ontologies are then translated into SQL scripts to perform the data integration on the database level.

The system works by describing both the source system (EMR) and the target dataset independently with ontologies. The source ontology is an abstract, technical description of the source system that provides information on how to access the data records behind each form component, done in a machine-interpretable way. The target ontology describes the collection of medical concepts to be made available in the i2b2 system, along with semantic features available in i2b2 (e.g. a mono-hierarchy of medical concepts, data type, medication and lab value ranges – where applicable). The target ontology is also used to automatically create the i2b2 ontology.

In order to link the source system (EMR) to the target system (i2b2) and to perform the export, a mapping ontology has to be created. This ontology contains manually created semantic relationships between the source and the target ontology, which can be processed by a software component to perform the data transfer. Relationships can either be „simple“ one-to-one or „complex“ mappings, with full data filtering and transformation rules. The latter ones are defined by using intermediate operation nodes, as illustrated with the ADD node on the image above. Intermediate nodes can also be cascaded into full expression trees to allow comprehensive data filtering and arbitrary data transformation. By sequentially processing these nodes in a correct order, SQL statements are automatically generated to perform the data transfer from the source systems to the target i2b2 database.

Useful information:

Edit the target (pivot) ontology with OntoEdit
Map the concepts from the target ontology to those of the source ontology (the data elements in the EMR)
Translate all ontologies to SQL to perform the automatic data integration