Building a model for disease classification integration in oncology, an approach based on the national cancer institute thesaurus

Table 3 Comparison with the SEER conversion program according to the tumor type (hematopoietic and solid tumors) and the number of branches of the diagnosis lattice that are identified for an ICD-10 code / ICD-O-3 combination

	All	Hematopoietic tumors	Solid tumors
	N=42260 (%)	N=14213 (%)	N=28047 (%)
Related in the model*	42260 (100.0)	14213 (100.0)	28047 (100.0)
More than 1 branch ^a	15234 (36.1)	9910 (69.7)	5324 (18.9)
Mappings rebuilt from the model**	17766 (42.0)	739 (5.2)	17027 (60.7)
Non unique mappings ^b	4886 (27.5)	333 (45.1)	4553 (26.7)

^*Related in the model means that there is at least a common diagnosis inside the model that is instantiated by both the ICD-10 code and the ICD-O-3 combination
^**Mappings rebuilt from the model corresponds to the mappings that we were able to rebuild automatically from the model
^aMore than 1 branch means that there is more than one branch of the diagnosis lattice that was instantiated by both the ICD-10 code and the ICD-O-3 combination
^bNon unique mappings means that the topography-morphology combination was also mapped to another ICD-10 code (inconsistent with the SEER file)

ISSN: 2041-1480