Vocabularies in the caBIG Federation

1
Vocabularies in the caBIG Federation

• George A. Komatsoulis, Ph.D
• Director, Quality Assurance and Compliance
• NCI Center for Biomedical Informatics and
  Information Technology (CBIIT)
• NCI Facilitator, caBIG Vocabulary and Common Data
  Element Workspace

2
Interoperability Defined

• ability of a system to access and use the parts
  or equipment of another system

Semanticinteroperability
Syntacticinteroperability
Courtesy Charlie Mead
3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
caBIG Compatibility Guidelines
8
Uses for controlled biomedical terminology

• Recording Experimental Data
• By using terminology whose meaning is agreed upon
  by all, meaning can be unambiguously conveyed
• Capturing Semantic Metadata
• The meaning of data is still affected by its
  context data interpretation is different if
  there is an observation of anemia as a disease
  symptom as opposed to an adverse event
• Computable Semantic Interoperability
• Provide a mechanism to describe an information
  system in such a way that two independently
  developed information systems can interoperate
• In caBIG, the Vocabulary and Common Data Element
  (VCDE) Workspace is responsible for semantic
  interoperability, and hence for vocabulary usage
  within the caBIG Federation.

9
Recording Experimental Data
10
Enterprise Vocabulary Services

• caBIG leverages the work of the NCI Enterprise
  Vocabulary Services (EVS)
• EVS provides vocabulary products and services to
  the NCI and the external scientific community
  including
• NCI Thesaurus
• NCI Metathesaurus
• External vocabularies such as LOINC, SNOMED, GO,
  etc.
• EVS provides access to these and other biomedical
  terminologies, in the form of human accessible
  tools such as the NCI Terminology Browser, and in
  the form of computer accessible Application
  Programming Interfaces (APIs)

11
caBIG Vocabulary Development

• Generally speaking, the caBIG program will adopt
  existing terminologies before funding the
  development of new terminologies
• For example, the Gene Ontology is considered a
  caBIG approved terminology
• caBIG has funded the development of three
  terminology products where the community
  indicated that gaps existed in current
  vocabularies
• Nutrition Terminology Lynne Wilkins (CRCH)
• RadLex playbook Curt Langlotz (RSNA)
• Mouse/Human Anatomy Map Martin Ringwald (JAX)

12
What makes a good biomedical terminology?

• Concept orientation
• Community acceptance
• Appropriate QA/QC
• English language definitions for all concepts
• Comprehensive documentation
• Easy accessibility and/or distribution
• Appropriate editorial policies/controls (change
  management)
• Good Usability/Reproducibility/Understandability
  (URU)
• Freedom from IP or licensing Constraints
• Regulatory or Sponsor requirements

13
Practical Considerations for Evaluating
Terminologies

• The guidelines adopted by the VCDE workspace
  described ideals not practical tests of
  vocabulary goodness of fit.
• The VCDE Workspace in cooperation with NCI EVS
  and the National Library of Medicine) engaged in
  three experiments to devise an effective strategy
  for evaluating a vocabulary against caBIG
  standards.
• NCI Thesaurus (NCIt) Olivier Bodenreider (NLM)
• Gene Ontology (GO) Martin Ringwald (Jackson
  Labs)
• Common Toxicity Criteria Adverse Events (CTCAE)
  James Cimino (Columbia)
• NCIt and GO have been designated caBIG approved
  vocabularies CTCAE comments have been sent to
  the owner of the vocabulary (NCI Cancer
  Therapeutics Evaluation Program).
• Standardized evaluation protocols are being
  derived from these experiments.

14
Development and Distribution of Terminologies

• caBIG is a federation, and this federation must
  extend to the development and distribution of
  terminology.
• Federated terminology development presents many
  challenges caBIG and EVS has focused on three
  main technical challenges
• Editing The NCI EVS is developing new technology
  for community development of content for NCI (and
  other) terminology products, including a
  customized version of the Stanford Protégé tool.
• Identification The VCDE workspace is developing
  a distributed terminology ID system (based on
  LSID) that will uniquely identify terminology in
  a distributed environment
• Distribution caBIG has been funding the
  development of the LexBIG Open Source terminology
  server. There are currently two adopters, NCI EVS
  and UK cancerGrid

15
Capturing Semantic Metadata
16
Semantic Metadata

• As described earlier, it is important to capture
  both the data being recorded and the context
  around that particular data capture event.
• Specifically, it is necessary to capture the
  meaning of what we are recording and what
  constitutes a valid response in that context.
• An international standard (ISO/IEC 11179) exists
  for such semantic metadata, this standard has
  been leveraged to create the cancer Data
  Standards Repository (caDSR).
• The basic unit of metadata in the caDSR is a
  Common Data Element or CDE.

17
The ISO 11179 Model in the caDSR
18
caDSR cancer Data Standards Repository

• The caDSR is an enterprise class system with a
  series of tools for metadata management
• CDE Browser
• UML Model Browser
• Curation Tool
• Administration Tool
• Sentinel Tool
• Form Builder
• The caDSR supports full metadata lifecycle (from
  draft new to retired) and additional
  registration statuses for classifying CDEs
• The caDSR is enabling technology for Computable
  Semantic Interoperability

19
Computable Semantic Interoperability
20
Object Oriented Information Systems
21
Semantic Interoperability CDEs describing
Information Systems
22
caBIG Interoperability by CDE
23
Data Standards
24
NCI Data Standards Activities

• Within caBIGTM the VCDE workspace works with
  domain experts to facilitate the development of
  standard vocabularies and data elements
• The VCDE workspace uses a consensus driven
  approach that accepts input from stakeholders in
  the cancer research community, including NCI
  funded cancer centers, cooperative groups and
  other standards organizations.
• These data standards drive the design of
  information systems whose development is funded
  by the NCI, or that desire to become caBIG
  compatible.
• The NCI participates in a number of other data
  standards projects, including CDISC (including
  the CDASH initiative), HL7 and the BRIDG project
• Harmonization with BRIDG is required of all newly
  funded projects in the caBIG program
• The report of the NCI Clinical Trials Working
  Group (CTWG) has initiated a CRF Harmonization
  activity within the Study Conduct SIG of the
  caBIG Clinical Trials Management Systems
  workspace. This group is closely engaged with the
  CDASH initiative.

25
Acknowledgements

• NCI CBIIT
• Dr. Ken Buetow
• Dr. Peter Covitz
• Dr. Leslie Derr
• Dr. Mary Jo Deering
• Dr. Frank Hartel
• Dr. Sherri de Coronado
• Dr. Gilberto Fragoso
• Ms. Denise Warzel
• Mr. Avinash Shanbhag
• caBIG VCDE Workspace
• Dr. Brian Davis
• Dr. Michael Keller
• NCI EVS Team
• NCI CDE Content Team

• Mayo Clinic LexBIG Team
• Dr. James Buntrock
• Dr. Chris Chute
• Mr. Harold Solbrig
• RSNA RadLex Team
• Dr. Curt Langlotz
• CRCH Nutrition Ontology Team
• Dr. Lynne Wilkins
• Jackson Labs Anatomy Team
• Dr. Martin Ringwald
• caBIG VCDE Vocabulary Teams
• Dr. Martin Ringwald (Jackson Labs)
• Dr. Olivier Bodenreider (NLM)
• Dr. James Cimino (Columbia)
• caBIG VCDE Participants
• caBIG Patient Advocates