Example pathology report snippet:
A. Brain, right frontal mass, resection: Glioblastoma, WHO grade 4 Additional molecular tests have been ordered; results will be reported in an addendum and separate molecular reports See note
If we have an NLP pipeline that is extracting ICDO3 site and histology, then we will need to extract two concepts (Concept ID: 44497946 = 'Frontal lobe' and Concept ID: 44499207= 'Glioblastoma, NOS') and then pre-coordinate them to the concept Concept ID: 44500954 = 'Glioblastoma, NOS, of frontal lobe' and then map that concept to the standard concept Concept ID: 36717624 = 'Primary glioblastoma multiforme of frontal lobe'. But likely we will want to connect each post-coordinated, extracted concept to two different textual spans, as represented in the offset_start and offset_end columns. Indeed, likely we would prefer to have two entries in NOTE_NLP. Connected to each span offset. How do we want to handle this? Should we amend the ETL conventions to cover the case of it being possible for multiple entries in NOTE_NLP being connected to a clinical event to cover the case of pre-coordinating concepts?
Example pathology report snippet:
A. Brain, right frontal mass, resection: Glioblastoma, WHO grade 4 Additional molecular tests have been ordered; results will be reported in an addendum and separate molecular reports See noteIf we have an NLP pipeline that is extracting ICDO3 site and histology, then we will need to extract two concepts (Concept ID: 44497946 = 'Frontal lobe' and Concept ID: 44499207= 'Glioblastoma, NOS') and then pre-coordinate them to the concept Concept ID: 44500954 = 'Glioblastoma, NOS, of frontal lobe' and then map that concept to the standard concept Concept ID: 36717624 = 'Primary glioblastoma multiforme of frontal lobe'. But likely we will want to connect each post-coordinated, extracted concept to two different textual spans, as represented in the offset_start and offset_end columns. Indeed, likely we would prefer to have two entries in NOTE_NLP. Connected to each span offset. How do we want to handle this? Should we amend the ETL conventions to cover the case of it being possible for multiple entries in NOTE_NLP being connected to a clinical event to cover the case of pre-coordinating concepts?