Table 2.

Example of methods used for the extraction of cancer notifications data.

Method	Notifications Data	Example
Queensland Cancer Registry coding rules (including special casings)	Histological Type	Select the highest morphology if more than one morphology is stated.
	Histological Grade	Assign the highest grade or differentiation code.
	Primary Site	Code all leukaemia except myeloid sarcoma (M-99303) to C42.1 (bone marrow).
Domain knowledge	Primary Site	List of one-to-one only ICD-O morphology to site mappings.
SNOMED CT property access	Histological Type	Restrict SNOMED CT concepts to those with a ‘morphologic abnormality’ semantic category and those that have alternate terms with the following regular expression “M-[0–9]{5}”.
	Primary Site	Restrict SNOMED CT concepts to those with a ‘body structure’ semantic category.
SNOMED CT to ICD-O topography cross-maps	Primary Site	Map SNOMED CT ‘body structure’ concepts to ICD-O topography codes.
SNOMED CT Subsumption querying	Histological Type	Candidate ‘leukaemia’ concepts are found by testing subsumption by the ‘128931003 | Leukemia – category’ concept.
SNOMED CT Concept relationship querying	Primary Site	“Procedure site – Direct” and “Finding site” relationship values from concepts are used as candidate sites.
SNOMED CT querying using ad-hoc term expansion	Histological Type	The histological type and grade’s preferred terms were used to search for a more specific concept. For example, the query for “Follicular Lymphoma” + “Grade 3” would return the histological type M-96983, which is “Follicular Lymphoma, Grade 3”.
Relation extraction	Basis of Diagnosis	Identification of multiple concepts or terms within a search scope such as metastasis and lymph nodes within a sentence.
Keyword/phrase spotting	Histological Grade	Detect keywords or phrases that were unable to be (or unreliably) mapped to SNOMED CT. For example “poorly to moderately differentiated”.