Knowledge Representation, Structuring and Ontology Development

1
Knowledge Representation, Structuring and
Ontology Development

• Arden Homestead
• June 4-5, 2003

2
Knowledge-Based Terminology

• History
• Theory
• Principles
• Reality
• Other efforts
• Additional applications
• Additional research

3
History

• DXplain Juvenile Polyarteritis Nodusm and
  Kawasaki Disease?
• Comparing disease descriptions to detect
  redundancy
• Detecting redundancy in the COSTAR Directory
• UMLS
• CPMC CIS Terminology

4
Theory

1. Knowledge-based approach should make high-quality
   terminologies easier to create
2. Knowledge-based approach should make
   terminologies easier to use
3. Knowledge-based approach should make
   terminologies easier to maintain

5
Principles

• Concept-oriented
• Multiple hierarchies
• Semantic relations
• Inheritance

6
Reality The MED

• Local terminologies
• National standards
• UMLS
• Other projects
• 77,000 concepts
• Lab tests and drugs have definitions

7
Reality Use of the MED

• Translating local codes for storage in repository
• Display spreadsheets for summary reporting
• Class-based decision support
• Translation for information retrieval

8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Problem-Oriented Views of the EMR
Heart
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
Reality MED Maintenance

• Semi-automated lab update
• Automated pharmacy update
• Manual updates (ASV)
• Some audits done routinely

16
Other efforts

• SNOMED
• Galen
• UMLS
• KIF
• Conceptual graphs

17
Additional applications

• Infobutton Manager
• Cornell neurobiology
• At CPMC?

18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
Towards a ColumbiaCornell Collaboration

• Laboratory of Neuroinformatics
• Weill-Cornell campus
• Funded by the NIHs Human Brain Project
• We seek collaborators to help develop BrainML
• Data description language for experimental
  neurobiology
• XML syntax, XML Schema lexical structure
• Lexicon trees codify values for neuroscience
  attributes
• Built on our BrainMetaL metalanguage
• Not a single note but a suite

22
A Sample BrainML Lexicon
23
Additional research

• Additional representations (Protégé)
• Knowledege-based editing
• Local Updates
• Visualization
• Richer semantics