Imaging studies are often accompanied by a clinical report of findings. This is typically stored in a 'Structured Report' (SR). The text stored in these reports can be very helpful for cohort building. The free text and tree structure of SRs make them difficult to process into a relational database. A better approach is to use a dedicated NLP framework and query it from the Cohort Builder. One such NLP tool is SemEHR
RDMP supports cohort building plugins. A plugin has been created to interface with SR data held in a SemEHR instance.
Before you can use the plugin you must first load your data into a SemEHR instance.
https://github.com/SMI/StructuredReports/tree/main/src/applications/SRAnonTool
The REST API must be configured and accessible over HTTPS. Test that you are able to get a response from the API with your web browser
In RDMP API endpoints are modelled as Catalogue objects in the same way as regular relational database datasets. Create a new one by right clicking any existing Catalogue and selecting 'Create New Sem EHR Catalogue':
Alternatively from the command line you can run:
./rdmp CreateNewSemEHRCatalogue
Create a new CohortIdentificationConfiguration. Ensure you have enabled query caching.
Add your new 'Sem EHR Catalogue' to the root container of the configuration
Double click the added cohort set to launch the configuration screen.
Enter a search term, negation (optional), date range (optional) and Modality.
Ensure the Settings are correct and run the query. If the 'Working' column changes to 'Crashed' then you can double click the row to view the error:
To get back to the settings/query screen after error right click the cohort set and choose 'Edit'