Processing of confidential information from patient files

Important patient data in previous patient files can often only be found in unstructured doctor's letters, operation or pathology reports. With the help of automatic extraction methods, these data can be extracted and converted into a structured format.

In the cloud4health project, SCAI has set up generic text mining workflows for the extraction of medical text documents such as doctor's letters, operation or pathology reports. Examples of  workflows established by SCAI are the extraction of tumor classifications from pathology data, the retrospective construction of an endoprosthesis register, or the extraction of diagnosis and medication information. In the project IDSN a workflow for the information extraction of the evaluations of cognitive tests from physician documentations is established. This data will then be linked with further research data and evaluated in combination (see figure).

For many of the medical data, a high level of data protection is required both before and during the analysis process.  Fraunhofer SCAI, together with its clinical partners in cloud4health, has designed and implemented a data protection-compliant architecture that allows secure cloud-based processing of data.