Download - Supporting Non-NLP Experts in Creating Annotation Schemas for Extraction of Clinical Concepts
Supporting Non-NLP Experts in Creating Annotation Schemas for Extraction of Clinical Concepts
Wendy W. Chapman, PhD
Department of Biomedical InformaticsUniversity of Utah
Motivation
Common information models
– Enable interoperability
– Guide non-NLP experts in developing NLP apps
Creating Domain Knowledge Bases for NLP
User Domain Knowledge Base
User Domain Knowledge Base Schema OntologySchema Ontology
Modifier OntologyModifier OntologyDomain
Ontologies
Knowledge Author
Creating Domain Knowledge Bases for NLP
User Domain Knowledge Base
User Domain Knowledge Base Schema OntologySchema Ontology
Modifier OntologyModifier OntologyDomain
Ontologies
Knowledge Author
Schema Ontology
Semantic types (from cTAKES type system)– Allergy– Disease/Disorder– Encounter– Finding– Medication– Procedure/Intervention– Research Activity– Sign/ Symptom– Social History– Test
Schema Ontology: Elements
Schema Ontology: Relationships
Creating Domain Knowledge Bases for NLP
User Domain Knowledge Base
User Domain Knowledge Base Schema OntologySchema Ontology
Modifier OntologyModifier OntologyDomain
Ontologies
Knowledge Author
Modifier Ontology
Modifiers are important for interpreting text– Chest radiograph confirms pneumonia– Family history of pneumonia– No evidence of pneumonia
Allowable modifiersFor each clinical element
Modifier Ontology
Semantic Modifiers
• Dosage• Duration• Form• Frequency• Route• Status Change• Anatomic location• …
Linguistic Modifiers
• Negation• Uncertainty• Conditional• Experiencer• Future• Historical
Modifier Ontology
• Value Sets– NegEx and ConText lexicon– Value sets from existing
ontologies/vocabularies
• Labels for different languages– English– Swedish– German– French
ACTION: Forward CLOSURE: Because_Group BackwardPatient_GroupBidirectional But_GroupTerminate SecondaryTo_Group
Which_Group
CATEGORY: DefiniteExistence ITEMS: AlthoughDefiniteNegativeExistence As a part fromEquivocality As a cause for Experiencer (n
~400)FutureHistoricalIndicationProbableExistenceProbableNegatedExistencePseudoExperiencerPseudoFuturePseudoHistoricalPseudoNegation
ConText Elements
Modifier OntologyTypes of modifiersTypes of modifiers Linguistic
expressionsLinguistic
expressions
ActionsActions
TranslationsTranslations
Modifier Ontology: Elements
Schema Ontology Imports Modifier Ontology
Medications– Type– Dose– Frequency– Route
Diagnosis– Negation– Uncertainty– Severity– History– Experiencer
Creating Domain Knowledge Bases for NLP
User Domain Knowledge Base
User Domain Knowledge Base Schema OntologySchema Ontology
Modifier OntologyModifier OntologyDomain
Ontologies
Knowledge Author
Domain Knowledge Base for NLP
• Instance of schema ontology
• Clinical elements from a particular domain
Synonyms Misspellings
Regular expressions
Synonyms Misspellings
Regular expressions
Creating Domain Knowledge Bases for NLP
User Domain Knowledge Base
User Domain Knowledge Base Schema OntologySchema Ontology
Modifier OntologyModifier OntologyDomain
Ontologies
Knowledge Author
Map NLP Output to User KB
User’s ConceptsCough
DyspneaInfiltrate on CXR
WheezingFever
Cervical Lymphadenopathy
User’s ConceptsCough
DyspneaInfiltrate on CXR
WheezingFever
Cervical Lymphadenopathy
Controlled Vocabs
Dry cough Productive coughCoughHacking coughBloody cough
Controlled Vocabs
Dry cough Productive coughCoughHacking coughBloody cough
Which concepts?
User’s ConceptsCough
DyspneaInfiltrate on CXR
WheezingFever
Cervical Lymphadenopathy
User’s ConceptsCough
DyspneaInfiltrate on CXR
WheezingFever
Cervical Lymphadenopathy
Attribute-values
Temp 38.0CLow-grade temperature
Attribute-values
Temp 38.0CLow-grade temperature
What values?
Knowledge AuthorKnowledge Author
Disease: colon cancerExperiencer: familyNegation: noHistorical: yes
Disease: colon cancerExperiencer: familyNegation: noHistorical: yes
“Family history of colon cancer”
NLP Schema Domain Ontology
Knowledge Author
• Front end interface for users
• Back end– Schema ontology– Modifier ontology
• Output– Domain ontology– Schema for NLP system
B Scuba, F Fana, Liqin Wang, Mingyuan Zhang, Y Liu, M Kong, F Drews
Questions | Discussion
African American Adult
Ibuprofen
Ibuprofen p.o.
No family history of colon cancer
Linguistic modifiers
Suggests synonyms
Acknowledgments
• Lee Christensen• Melissa Tharp• Mike Conway• Danielle Mowery• Bill Scuba• Milan Kovacevich• Dieter Hillert• Samir Abdelrahman• Leah Willis• Bob Angell
• Harry Hochheiser
• Jan Wiebe
• Rebecca Hwa
• Guergana Savova
• Noemie Elhadad
• Michael Matheny
• Rob El-Kareh
• Ruth Reeves
• Qing Zeng
• Guy Divita
• Frank Drews
BLU Lab Collaborators• Sumithra Vellupilai• Maria Kvist• Maria Skeppstedt• Aron Henrikkson• Brian Chapman• David Carrell• Sascha Dublin• Zia Agha• Stephane Meystre• Scott DuVall• Jianlin Shi