Computer Representation of Adverse Events Following Immunizations Using
- Slides: 21
Computer Representation of Adverse Events Following Immunizations Using Semantic Web Technology Herman Tolentino, MD VAU / BVPDB / ESD / NIP / CDC Public Health Informatics Fellow, PHIFP March 24, 2004
Outline o Introduction o Methods o Results o Conclusion
Semantic Web • A mesh of information • Uses a form of XML called Resource Description Framework (RDF) to store knowledge • Readable by machines • Knowledge is organized as ontologies.
What is an ontology? • A formal, structured conceptualization of a body of knowledge = what we know about something (e. g. , adverse events) • Knowledge base = how we represent the ontology in a system • Our body of knowledge is adverse events following immunization (AEFI)
Seizure Loss of consciousness History Witnessed Manifestations motor manifestations Tonic Clonic Tonic-clonic Level 1 Level 2 Diagnostic certainty
Uses of Ontologies in Public Health • Semantic interoperability • Semantic enhancement • Knowledge repository
Uses of Ontologies in Public Health Semantic Interoperability • Exchange of data at the level of meaning • Semantic problem example: fundus, pyogenic granuloma • Information systems understand each other and allow for flexible retrieval of information
Uses of Ontologies in Public Health Semantic Enhancement • Getting to all possible representations of a controlled vocabulary code • By searching for related concepts in the Unified Medical Language System (UMLS) • Hepatitis - related concepts would be jaundice, icterus, hepatomegaly
Uses of Ontologies in Public Health Knowledge Repository • Knowledge explosion: If a physician reads only 2 journals a day he will be behind by 800 years worth of knowledge at the end of one year (Koop, 1999). • Enabled by rapid advances in and decreasing costs of technology
Adverse Event Ontology Project • OVAE (Ontology for Vaccine Adverse Events) • Problems addressed: • Digitally represent standardized case definitions (semantic interoperability) • Cast a wider net for cases in surveillance systems by getting to related codes using the ontology and UMLS (semantic enhancement) • Knowledge management (repository)
Framework Conceptual Model Case Definitions RDF UMLS RDF Phase I Adverse Events Following Immunization Ontology Phase II User Interface Creation and Inferencing Phase III
Results o Conceptual model of adverse events (BC) o Translation of case definitions and UMLS concepts to Resource Description Framework (RDF) format
1 ADVERSE EVENTS 2 CASE DEFINITIONS 3 CONTROLLED VOCABULARIES 4 CLINICAL MANIFESTATIONS 5 LEVEL OF DIAGNOSTIC CERTAINTY
Seizure Case Definition • Level 1 Diagnostic Certainty • Witnessed sudden loss of consciousness AND • Generalized, tonic, clonic, tonic-clonic, OR atonic motor manifestations • Level 2 Diagnostic Certainty • History of unconsciousness AND • Generalized, tonic, clonic, tonic-clonic, OR atonic motor manifestations
How the UMLS is Used • UMLS concepts are identified using Concept Unique Identifiers (CUIs). • Controlled vocabulary codes are related to concepts. • Using the UMLS and the adverse event ontology, we can semantically enhance event detection by linking concepts in the ontology to concepts in the UMLS.
RDF file of a UMLS table
Seizure Loss of consciousness Concept Unique Identifier (CUI) Manifestations C 000001 Witnessed ICD 9 concepts C 000002 Motor manifestations Tonic Clonic History UMLS Level 1 codes Level 2 ICD 9 Tonic-clonic Diagnostic certainty
Conclusions • The domain of ontology creation is an emerging field in health care. • Advances in technology permit creation of largescale ontologies. • We can use ontologies for semantic interoperability and semantic enhancement and as knowledge repositories. • Collaboration leverages partners’ resources and fosters shared learning.
Future Directions • Phase 1: Continue ontology development • Phase 2: Merge two conceptual domains: vaccine adverse events and UMLS. • Phase 3: Create query interface for inferencing engine.
Acknowledgements • Brighton Collaboration - a voluntary organization of international experts that develops standardized case definitions for AEFI • Office of High Performance Computing and Communications of the National Library of Medicine
Correspondence Herman Tolentino, MD HTolentino@CDC. gov Daniel Payne, MSPH, Ph. D DPayne@CDC. gov BVPDB / ESD / NIP / CDC 1600 Clifton Road NE, Mailstop E-61 Atlanta, GA 30329
- Aefi examples
- Adverse events in hospital
- Adverse events in hospital
- Puerperal sepsis meaning
- Georgia registry of immunizations
- Links louisiana immunizations
- Mutually exclusive vs non mutually exclusive
- Polynomial representation using arrays
- Polynomial division using linked list in c
- Data representation computer science
- Data representation in computer organization
- Boundary representation in computer graphics
- Data representation in computer architecture
- Information representation in computer
- Sweep representation in computer graphics
- Data representation and computer arithmetic
- Data types and representation
- Data representation in computer organization
- Float number
- Hex code ascii
- Surface representation in computer graphics
- Adverse reaction definition