Language, medical terminologies and structured electronic patient records: how to escape the Bermuda triangle.

1
Language, medical terminologies and structured
electronic patient records how to escape the
Bermuda triangle.

• Dr. W. Ceusters
• Dir RD
• Language Computing nv

2
The Medical Informatics dogma

• To structure or NOT to be
• Fact computers can only deal with a structured
  representation of reality
• structured data
• relational databases, spread sheets
• structured information
• XML simulates context
• structured knowledge
• rule-based knowledge systems
• Conclusion a need for structured data entry

3
Structured data entry

• Current technical solutions
• rigid data entry forms
• coding and classification systems
• But
• the description of biological variability
  requires the flexibility of natural language and
  it is generally desirable not to interfere with
  the traditional manner of medical recording
  (Wiederhold, 1980)
• Initiatives to facilitate the entry of narrative
  data have focused on the control rather than the
  ease of data entry (Tanghe, 1997)

4
Drawbacks of structured data entry

• Loss of information
• qualitatively
• limited expressiviness of coding and
  classification systems, controled vocabularies,
  and traditional medical terminologies
• use of purpose oriented systems
• dont use data for another purpose than
  originally foreseen (J VDL)
• quantitatively
• to time-consuming to code all information
  manually
• Speech recognition and structured data entry
  forms are not best friends

5
The three pilars of modern M.I.

• Clinical language
• medical narrative

• Clinical terminologies
• coding and classification systems
• nomenclatures
• formal ontologies

• Electronic Healthcare Record Systems

6
How to harmonise the pilars ?
Domain of discourse healthcare
Comparability of data Crossborder care Decision
support Abstraction / grouping ...
Individual patient care Seemless care Historical
overview ...
Faithfull data recording Sufficient level of
detail ...
7
Six possible approaches
8
What do the symbols stand for ?

• Text based EHCRS able to generate structured data
  
• An EHCR exclusively build around a collection of
  coded data generated out of free text
• A multimedia EHCRS with clinical narrative
  registration and structured data generation
• A multimedia EHCRS with structured data entry and
  text generation
• An EHCR exclusively build around texts generated
  out of controled vocabularies
• An EHCR exclusively build around a collection of
  structured data able to generate text

9
Main harmonisation principle

• Accept what is offered and offer what you
  accepted...

10
Harmony maximum total surface
100
73
43
59
41
83
11
Terminology often hurts !

• The more dominant the position of terminology,
  the more is lost

• by offering too little to the component over
  which it dominates

12
Terminology centered approaches
13
What about the UMLS approach ?

• Simply using everything in the Metathesaurus does
  not make a good coding system W. Hole,
  2000
• The problems with the Metathesaurus as a single
  monolithic vocabulary are
• There is a wide range of granularity of terms in
  different vocabularies
• The Metathesaurus itself has no unifying
  hierarchy
• There may be other features of vocabularies that
  get lost in their "homogenisation" upon being
  entered into the Metathesaurus. ?W.
  Hersh, 2000?

14
Are formal ontologies better ?

• The current implementation of SNOMED-RT does not
  have the depth of semantics necessary to arrive
  at comparable data or to algorithmically map to
  classifications such as ICD-9-CM
• ?Peter Elkine, 1999?
• A serious limitation of the Galen approach is
  that specialisation is invariably linked to a
  conceptual relation
• Udo Hahn, 1999

15
What is the problem ?

• ICD, ICPC, MedDRA,
• too purpose oriented
• UMLS
• build without a formal ontology
• Galen, SNOMED-RT,
• build without language (as a medium of
  communication) in mind

16
The future is in linguistic ontologies
Formal Domain Ontology
Linguistic Ontology
MedDRA
17
Linguistic ontologies in action
WE
WE-P-Type
WE-P-State
P-P-Type
P-P-State
Material Entity
4
Feature
Feature State
P-Scale
Scale State
1
Temperature State
Water
3
High Scale State
5
Temperature
high
High Temperature State
temperature
6
Warm water
1 H-P-T 2 H-WE-S 3 H-WE-P-S 4
H-P-P-S 5 H-Expr-P-S 6 H-Real-P-S
2
warm
Warm water
18
The problem summarised

• natural language is the only medium that is able
  to communicate clinical information about
  individual patients without loss of necessary
  detail
• structured data repositories are required to make
  subsequent analyses possible
• any transformation from free language to coding
  and classification systems results in information
  loss that is unacceptable for individual patient
  care, but at the other hand is a conditio sine
  qua non for population based studies
• todays graphical user interfaces can deal
  reasonably well with picking lists build around
  controlled vocabularies that fulfil a bridging
  function from free language towards coding and
  classification systems but are incompatible with
  speech recognition technology.

19
Towards an adequate solution
20
Questions ?