Achieving Semantic Interoperability of Cancer Registries

1
Achieving Semantic Interoperability of Cancer
Registries

• NAACCR 2007 Annual Meeting
• June 7, 2007
• Peter Winkelstein, MD
• Chief, Division of General Pediatrics
• CMIO, UPA and UBA
• University at Buffalo, New York
• pwink_at_buffalo.edu

2
Informatics Goals for Cancer Registries

• Improve efficiency and accuracy of data entry
• Automated data extraction from EHRs and other
  systems
• Automated data checking
• Improve data extraction
• Facilitate reporting (querying)
• Facilitate secondary use of registry data
• Translational research
• Data mining

3
Computable Health Information

• Registry data must be
• Stored in such a way as to be amenable to
  computational processes (e.g. database elements)
• Have carefully defined meaning (e.g. ontologies)
• Required for interoperability
• Required for robust querying

4
What is Ontology (philosophically)?

• A representation of some pre-existing domain of
  reality which
• reflects the properties of the objects within its
  domain in such a way that there is a systematic
  correlation between reality and the
  representation itself
• is intelligible to a domain expert
• is formalized in a way that allows it to support
  automatic information processing

(Source Werner Ceusters, 12/23/2005)
5
What is an ontology (practically)?

• A formal model of a domain which consists of
• Classes (aka concepts)
• Instances (particular members of classes)
• Instances are not generally part of published
  ontologies
• Relationships between classes (aka roles)
• In contrast to controlled vocabularies, which are
  word lists with no (or limited) relationships

6
Example
7
Syntactic vs. Semantic Interoperability

• The ability to exchange information among systems
  and software, especially EHRs, is called
  interoperability
• The ability to exchange data is called syntactic
  interoperability
• Interfaces/Messaging
• The ability to exchange data and meaning is
  called semantic interoperability
• Interfaces/Messaging plus Ontologies

8
Syntactic Interoperability
9
Semantic Interoperability
10
Syntactic Interoperability Alone Is Not Enough

• Free text data has little (if any)
  machine-interpretable meaning
• Data elements are not guaranteed interoperable
• Multiple encodings
• Inflexible encoding
• Incompatible lists/Granularity
• Data abstraction
• Impression vs. criteria
• Qualitative qualifiers
• Lack of standards

11
Semantic Interoperability

• Semantic interoperability allows the exchange of
  both structure and meaning among systems
• Requires data in structured form (database
  elements) plus encoding data with
  machine-interpretable meaning
• Ontologies are the tools for this encoding

12
State-of-the-art in Ontology

• There are few, if any, well-formed medical
  ontologies (or even controlled vocabularies)
• Notably for cancer informatics, the current state
  of the NCI Thesaurus is suboptimal

13
Some Existing Vocabularies and Ontologies

• Controlled Vocabularies
• DICOM
• LOINC
• HCPCS
• ICD-9
• CPT
• ICD-O
• MedDRA

• Ontologies
• HL7 RIM
• UMLS
• NCIT
• FMA
• GO
• SNOMED-CT

14
Examples

• NAACCR
• Standards for Cancer Registries
• Volume II
• Data Standards and Data Dictionary
• Eleventh Edition
• Record Layout Version 11.1

15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Keys to Achieving Semantic Interoperability

• Structured data collection (no free text)
• Highest granularity possible
• Careful definition of meaning for each data
  element
• Consistent use of each data element
• Assignment of standard code to each data element
  if possible
• Vocabulary creation and maintenance for meanings
  not available among current standards
• Publication and versioning of all definitions

27
Recommendations

• Cancer registries should move towards semantic
  interoperability
• This would facilitate improvements in both data
  collection and extraction
• A strong link to EHR data is particularly
  desirable
• Registry organizations should actively work with
  vendors towards semantic interoperability with
  oncology EHRs