6. Index
• Types of language
• Semantic resources
• How to deal with this semantic ecosystem?
603/07/2014
7. Types of language
• There are different types of language
according to their use in Information Systems
(IS):
– Natural language (can rises up to 90%).
– Controlled vocabularies (the remaining 10%):
• Interface (input) vocabularies.
• Reference vocabularies.
• Aggregation (output) vocabularies.
703/07/2014
8. Types of language
• Natural language
– It refers to both oral and written language used by
professionals to communicate with patients and
other physicians:
• It is the most friendly, it is natural.
– Depends on the context and interlocutors.
– It includes jargon, acronyms, homonyms, etc.
– Being very reach and flexible.
– It can be very specialized but also very ambiguous.
803/07/2014
9. Types of language
• Natural language is human oriented:
– It is not the best choice for IS to advice and
support healthcare professionals.
• It is necessary to normalize the relevant
information according to our objective:
– PHR interoperability: Exchange and query?
– Big Data analysis: Plan and discover?
– Healthcare System: Control, make
recommendations and prevention?
– Upcoming: Research and Innovation?
903/07/2014
10. Types of language
• Controlled vocabularies are normalized
languages that can be understand by IS, as they
can:
– Identify, localize and differentiate its elements.
– Automatically process it.
– Have a representation of the meanings without
ambiguity and using different levels of detail.
• This normalization involves using codes and
attributes to represent specific ideas or units of
meaning.
1003/07/2014
11. Types of language
• There are 3 types of controlled vocabulary:
– Interface (input):
• Vocabulary used directly by healthcare professionals to
record the information of patients in IS.
• It represents the first level of formalization and concretion.
– Reference:
• It allows relating the vocabularies in order to compare and
consider information from different sources.
• It is granular and presents the maximum level of detail.
– Aggregation (output):
• Objective: exploiting different kind of information like:
statistics, plan, reimbursement, population analysis, etc.
• Interpretation from different points of view is allowed
(manager and financial).
1103/07/2014
12. Semantic resources
• Semantic resources are controlled vocabularies
used to represent, to codify, to index and/or to
mark data, information or content.
• There are many types of resources according to
their structure and objectives:
– Catalogues
– Thesaurus
– Classifications
– Terminologies
– Ontologies
– …
1203/07/2014
13. Semantic resources
• Catalogues
– Being plain lists of codes and descriptions
normally close to a limited domain.
– They can also have other few attributes.
– Having usually ONE description per code.
– They are often local but they can be international
as well.
– Objective: to facilitate the identification and
localization of their already sorted elements.
1303/07/2014
15. Semantic resources
• Thesaurus
– They are formalized lists of terms related to each
other through hierarchy, equivalency or
associative links:
• For example synonyms of the same idea.
– Objective: to index or mark content in order to
facilitate the recovery of documents, articles or
other publications in databases of documentation.
1503/07/2014
16. • Thesaurus
– Example: COSTART (The Coding Symbols for a
Thesaurus of Adverse Reaction Terms)
1603/07/2014
Semantic resources
17. Semantic resources
• Classifications
– They are made of elements with codes and
descriptions that can be grouped in chapters of
content.
– Having concepts organized by classes and subclasses,
following a tree structure (modelled by codes).
– They have different levels of detail (that can include
miscellaneous and unspecific concepts).
– They tend to be international but also can be local.
– Normally they are closed to only one single domain.
– Objective: To classify certain information of patients
(for example diagnosis or procedures).
1703/07/2014
19. Semantic resources
• Terminologies
– They are made of terms related to each other through
hierarchy and other kind of relationships.
– Their elements can have more than one description
and many other attributes.
– Normally they are international and provide a
standard to achieve semantic interoperability.
– Having high level of detail and granularity in the
representation of its elements.
– Objective: to relate concepts from different sources
and controlled vocabularies that represent the same
idea, using the maximum level of detail.
1903/07/2014
21. Semantic resources
• Ontologies
– Being semantic networks of related concepts
through many different kinds of associations.
– They have elements that can have many
descriptions and attributes.
– Representing one single domain normally.
– They have concepts with a high level of
abstraction:
• Each of them can be codified by a terminology.
– Objective: to represent meanings to allow its
automatic processing in IS.
2103/07/2014
26. 2603/07/2014
• Healthcare centres have multiple semantic resources for
each domain:
– Medicines, active ingredients, dose forms, routes of
administration, etc.
– Diagnosis, morphology, findings, topologies, etc.
– Other master data like genders, types of address, types of
professionals for roles, etc.
• And they need to map vocabularies to represent the
information using the maximum level of detail and to
communicate with other centres and organizations.
• Additionally, all these resources need to be maintained and
distributed to all the clinical work stations.
Semantic resources
28. How to deal with this semantic ecosystem?
2803/07/2014
• Using tools that have been specifically
designed to manage, distribute, develop and
explore controlled vocabularies:
– Terminology servers (for example HealthTerm and
ITServer).
• And using this tools directly at the clinical
work stations in the point of care by
healthcare professionals that is assisting a
patient:
– And recording directly the information using
standards.
29. How to deal with this semantic ecosystem?
2903/07/2014
https://www.youtube.com/watch?v=6v48LqvMk9c
31. Index
• SNOMED CT as reference terminology
• Introduction
• Functionalities
• Commercial Tools
– HealthTerm
– ITServer
• Clinical Work Station
• References
02/07/2014 31
32. SNOMED CT
• SNOMED CT is a multilingual and international terminology.
• SNOMED CT is a semantic standard that can represent elements of
many healthcare domains:
– Diseases, pharmaceutical products, routes of administration,
substances, procedures, etc.
• SNOMED CT uses concepts to represent ideas:
– Each concept has more than one description.
– Each concept is related to other concepts trough relationships
performed by many kind of associations (not only hierarchy).
• SNOMED CT also includes extensions that can be viewed as local
versions of the terminology but respecting and following the
standard.
02/07/2014 32
33. SNOMED CT
• SNOMED CT can be used as an interface and
aggregation vocabulary:
– It has accessible descriptions made BY and FOR
healthcare professionals.
– It includes the subset mechanism to facilitate its
use in/with IS and by clinicians:
• Capacity to group concepts, descriptions or
relationships in order to use only those we really need.
02/07/2014 33
34. SNOMED CT
• BUT… the real strength comes when we use
SNOMED CT as a reference terminology:
– It has relationships of many kinds:
• Has a multi-hierarchy structure that relates concepts of
different domains.
– It is granular and allows representation of
concepts using the maximum level of detail.
– And it includes a mapping mechanism used to
relate their concepts with elements of other
controlled vocabularies.
02/07/2014 34
35. Introduction
• We can found 3 types of tools that work with
controlled vocabularies:
– Browsers.
– Terminology servers.
– Exploiting tools.
02/07/2014 35
36. Introduction
• Browsers
– They allow us to explore the content of controlled
vocabularies through:
• Searches: simple and advance.
• Navigation: plain list or hierarchy discovery.
– They display all the information about controlled
vocabularies: common name, version, edition,
author, etc.
• And also about its content, for each element they
show: ID or CODE, description(s), relationships and any
other attribute.
02/07/2014 36
37. Introduction
• Terminology servers
– Tools specially designed to allow users and
systems to work with controlled vocabularies:
• Maintenance, distribution, development, etc.
– That allow “terminology (to be used) as a service”.
– They are not databases managers.
02/07/2014 37
38. Introduction
• Exploiting tools
– They use controlled vocabularies with other main
proposes:
• Like to prescribe a medicine, to report a discharge of a
patient or to order laboratory test analysis.
– Clinical Work Stations of points of care are exploiting
tools.
– Healthcare professionals uses these tools to record
information in the IS of their centre:
• This information ideally will be stored in a structured format
and using standards…
– But professionals don’t need to see any code.
02/07/2014 38
39. Introduction
• These tools are not mutually exclusive:
– A terminology server should have a browser.
– And most of the exploiting tools have a (partial)
browser to help users to find terms.
• And can be used by persons and systems:
– Browsers and exploiting tools are used by persons.
– Terminology servers are used by persons and
systems.
02/07/2014 39
40. Functionalities
• Related to controlled vocabularies, their
components, subsets and mappings:
– Import in a standard or common format:
• First release and historical mechanism.
• Other metadata.
– Representation.
– Exploration with searches and navigation.
• Including queries.
– Export in a standard or common format.
02/07/2014 40
41. Functionalities
• Related to controlled vocabularies, their
components, subsets and mappings:
– Creation:
• Including translation.
– Edition.
– Inactivation.
02/07/2014 41
42. Functionalities
• Additional functionalities for SNOMED CT:
– Management of extensions:
• Includes operations to create, modify and translate
components of SNOMED CT.
– Support to post-coordination and management of
clinical expressions:
• Using more than one concept to represent an idea.
02/07/2014 42
43. Functionalities
• Terminology servers can offer these
functionalities in 2 modes:
– Through a user interface: for persons (terminology
managers).
– As a (web) service: for systems.
• And they have other functionalities as
management systems like:
– Control of versions.
– Management of users, rolls and profiles.
– …
02/07/2014 43
44. Functionalities
• The main functionalities of terminology
servers are described in CTS2 standard of HL7:
– Common Terminology Services Release 2.
– Its definition is at a functional level and establish
profiles and criteria of accomplishment.
– Its specification is independent of controlled
vocabularies, languages, systems and
technologies:
• Allowing different implementations approaches.
02/07/2014 44
48. Clinical Work Station
• How can we use controlled vocabularies
stored and managed from terminology servers
in Clinical Work Stations?
02/07/2014 48