SEMI-AUTOMATIC MAPPING OF HETEROGENEOUS PATHOLOGICAL CANCER DATA SOURCES TO HL7FHIR RESOURCES

Abstract: Heterogeneous health care data sources complicate the distribution of structured information between di‚fferent IT-systems involving cancer care in Sweden. Th‘is causes additional overhead for statisticians and system developers where they need to create tailored solutions for each problem. Th‘ere is a need for standardization of this data. HL7 FHIR is a standard for health care data exchange and can provide solutions for integration requirements in the form of so-called resources. Implementers must map their problems to resources and their content to do so. Mapping these data sources can be complex and time-consuming to do manually, but is there a way to do this in a semi-automatic way? ‘This thesis presents workƒflows including implementations utilizing syntactic and semantic textual similarity methods to semi-automatically map heterogeneous pathology datasets to a selection of structured HL7 FHIR resources. Th‘is is done by extracting di‚fferent textual representations that represent the meaning of each datasource so that syntactic and semantic textual similarity methods can be applied.‘ The workflƒows have shown to produce some relevant mapping alternatives if textual representations being compared are not too far from each other in terms of length and terminology used.