Presentation slides for the Medical Informatics Europe, 2014, paper: A Framework for Evaluating and Utilizing Medical Terminology Mappings

1. A Framework for Evaluating and Utilizing
Medical Terminology Mappings
EHR4CR – Open PHACTS, SALUS and W3C collaboration
Sajjad Hussain1, Hong Sun2, Ali Anil Sinaci3, Gokce Banu Laleci Erturkmen3, Charlie Mead4,
Alasdair Gray5, Deborah McGuinness6, Eric Prud’Hommeaux7, Christel Daniel1, Kerstin Forsberg8
MIE2014 2-Sept-2014
EHR4CR: 1INSERM UMRS 1142, Paris, France; 8 AstraZeneca, R&D Information, Mölndal Sweden
Open PHACTS: 5School of Mathematical and Computer Sciences, Heriot-Watt University
SALUS: 3Software Research, Development and Consultancy, Ankara, Turkey,
2Advanced Clinical Applications Research Group, Agfa HealthCare, Gent, Belgium
W3C: 4Health Care and Life Sciences IG, 7MIT, Cambridge, MA, USA,
6Department of Computer Science, Rensselaer Polytechnic Institute, Troy, US
1
2014 Medical Informatics Europe
Version 1.0 http://slideshare.net/kerfors/MIE2014

2. Objective
• Show the challenging nature of mapping
utilization among different terminologies.
• A framework built upon existing terminology
mappings to:
– Infer new mappings for different use cases.
– Present provenance of the mappings together
with the justification information.
– Perform mapping validation in order to show
that inferred mappings can be erroneous.
• Enable a more collaborative semantic landscape
with providers and consumers of terminology
mappings.
2
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

3. Semantic landscape 1(3)
3
For more information about these see the reference
slides in the end of this slide deck.
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014
Consumers and, somewhat
reluctant, creators of mappings

4. Semantic landscape 2(3)
4
Providers of terminology mappings,
some examples
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014
Consumers and, somewhat
reluctant, creators of mappings

5. Semantic landscape 3(3)
5
Providers of terminology mappings,
some examples
Providers of terminologies,
some examples
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014
Consumers and, somewhat
reluctant, creators of mappings

6. Rationale
• Challenging nature of mapping utilization, or
“How hard can it be?”
– Appear to the uninitiated as a simple exercise like “this
term in this terminology is the same as that term in that
terminology”
6
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

7. Example Scenario
• Challenging nature of mapping utilization, or
“How hard can it be?”
– Appear to the uninitiated as a simple exercise like “this
term in this terminology is the same as that term in that
terminology”
7

8. Example Scenario 1(3)
8
Defined
Mappings

9. Example Scenario 2(3)
9
matches
matches
matches
Defined
Mappings
Inferred
Mappings

10. Example Scenario 3(3)
10
matches
matches
matches
Defined
Mappings
Inferred
Mappings
matches
Problematic
Mappings

11. “It’s complicated”. So, we often become, somewhat
reluctant, creators of our own mappings
• Availability of up-to-date information to assess the suitability
of a given terminology for a particular use case.
• Difficulty of correctly using complex, rapidly evolving
terminologies.
• Differences in granularity between the source and target
terminologies.
• Lack of semantic mappings in order to completely and
unambiguously define computationally equivalent semantics.
• Lack of provenance information, i.e. how, when and for what
purposes the mappings were created.
• Time and effort required to complete and evaluate mappings.
11
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

12. Objective: A more collaborative
semantic landscape
12
Informed consumers of
terminology mappings
Value adding providers of
terminology mappings
Value adding providers of
terminologies

13. Framework
13
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

14. Mapping Strategies
• Lexical Mappings (LOOM) generated by performing lexical comparison between
preferred labels and alternative labels of terms. These mappings are represented via
skos:closeMatch property.
• Xref OBO Mappings Xref and Dbxref are properties used by ontology developers to
refer to an analogous term in another vocabulary. These mappings are represented
via skos:relatedMatch property.
• CUI Mappings from UMLS are extracted by utilizing the same Concept Unique
Identifier (CUI) annotation as join point of similar terms from different vocabularies.
These mappings are represented via skos:closeMatch property.
• URI-based Mappings are generated identity mappings between term concepts in
different ontologies that are represented by the same URI. These mappings are
represented via skos:exactMatch property.
14
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

15. Terminology Mappings Validation Schemes

16. Collaborative semantic landscape
16
Informed consumers of
terminology mappings
Value adding providers of
terminology mappings
Value adding providers of
terminologies
Enabled by applications of
the RDF standard

17. 17
Application of RDF for
representing mappings
Enabled by applications of
the RDF standard

18. 18
Application of RDF for
representing provenance
Enabled by applications of
the RDF standard

19. Applications of RDF for packaging assertions
19
(e.g. mappings) with provenance
Enabled by applications of
the RDF standard

20. 20
Applications of RDF for describing
datasets and linksets with justifications
Enabled by applications of
the RDF standard

21. Example Scenario
21
matches
matches
matches
Defined
Mappings
Inferred
Mappings
matches
Problematic
Mappings

22. 22
Example Scenario
matches
Defined
Mappings
Inferred
Mappings

23. matches
Defined
Mappings
Inferred
Mappings
23
SKOS/RDF for representing mappings
ICD9CM:999.4 skos:exactMatch SNOMEDCT:21332003
SNOMEDCT:21332003 skos:exactMatch MedDRA:10067113
ICD9CM:999.4 skos:exactMatch MedDRA:10067113

24. Nanopublication for packaging mappings
and mapping provenance representations
ICD9CM:999.4 skos:exactMatch MedDRA:10067113
matches
Defined
Mappings
Inferred
Mappings
24
Assertion
Justification trace generated from EYE reasoning engine

25. Justification Vocabulary terms for
Relating Terminology Concepts/Terms
25
??

26. 26
Applications of RDF for describing
datasets and linksets with justifications
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014
Enabled by applications of
the RDF standard

27. Linksets: Justification Vocabulary Terms 1(3)
27
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

28. Linksets: Justification Vocabulary Terms 2(3)
28
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

29. Linksets: Justification Vocabulary Terms 3(3)
29
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

30. CIM Workshop at ISWC2014 to discuss:
Justification Vocabulary terms for
Relating Terminology Concepts/Terms
30

31. Acknowledgments
• Session chair
• MIE2014 organizers
• SALUS team: Hong Sun, Ali Anil Sinaci, Gokce Banu Laleci Erturkmen
– Support from the European Community’s Seventh Framework Programme
(FP7/2007–2013) under Grant Agreement No. ICT-287800, SALUS Project
(Scalable, Standard based Interoperability Framework for Sustainable
Proactive Post Market Safety Studies).
• EHR4CR team: WP4, WPG2, WP7 members
– Support from the Innovative Medicines Initiative Joint Undertaking under
grant agreement n° [No 115189]. European Union's Seventh Framework
Programme (FP7/2007-2013) and EFPIA companies
• Open PHACTS team: Alasdair Gray
• W3C HCLS team: Eric Prud’Hommeaux, Charlie Mead
31
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

32. Reference material
• Projects/organisations of the authors of this paper
• Example
– Mapping Representation using SKOS
– Mapping Provenance Representation
2014 Joint Summits on Translational Science 32

33. EHR4CR
Electronic Healthcare Record For Clinical Research
http://www.ehr4cr.eu/
• IMI (Innovative Medicine Initiative)
– Public-Private Partnership between EU and EFPIA
• ICT platform: using EHR data for supporting
clinical research
• Protocol feasibility
• Patient recruitment
• Clinical trial execution: Clinical Research Forms (eCRF)/
Individual Case Safety Reports (ICSR) prepopulation
• 33 European academic and industrial partners
– 11 pilot sites from 5 countries – 4 millions patients
33
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

34. Open PHACTS
Open Pharmacology Space
http://www.openphacts.org/
• IMI (Innovative Medicine Initiative)
• 31 partners: 10 pharma – 21 academic / SME
• The Challenge - Open standards for
drug discovery data
– Develop robust standards for solid
integration between data sources via
semantic technologies
– Implement the standards in a semantic
integration hub (“Open Pharmacological Space”)
– Deliver services to support on-going drug
discovery programs in pharma and
public domain
34
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

35. SALUS
Sustainable Proactive Post Market Safety Studies
http://www.salusproject.eu/
• European Commission (STREP)
• ICT platform : using EHRs data to improve post-market
safety activities on a proactive basis
• Semi-automatic notification of suspected adverse events
• Reporting adverse events (Individual Case Safety Reports
(ICSR) prepopulation)
• Post Marketing safety studies
• 8 European academic and industrial partners
– 2 pilot sites
• Lombardia Region (Italy) and Eastern Saxony (Germany)
35
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

36. W3C
Semantic Web Health Care and Life Sciences Interest
Group (HCLS IG)
http://www.w3.org/2001/sw/
• ..
36
2014 Medical Informatics Europe
http://slideshare.net/kerfors/MIE2014

37. Mapping Representation using SKOS
37
<http://purl.bioontology.org/ontology/ICD9CM/999.4>
<http://www.w3.org/2004/02/skos/core#broadMatch>
<http://purl.bioontology.org/ontology/MDR/10002198>,
<http://purl.bioontology.org/ontology/MDR/10002199>,
<http://purl.bioontology.org/ontology/MDR/10020751>,
<http://purl.bioontology.org/ontology/MDR/10067484> .
<http://purl.bioontology.org/ontology/MDR/10002198> a <http://www.w3.org/2004/02/skos/core#Concept>;
<http://purl.bioontology.org/ontology/MDR/level> "PT";
<http://www.w3.org/2004/02/skos/core#broader> <http://purl.bioontology.org/ontology/MDR/10002220>,
<http://purl.bioontology.org/ontology/MDR/10057181>;
<http://www.w3.org/2004/02/skos/core#inScheme> <http://purl.bioontology.org/ontology/MDR>;
<http://www.w3.org/2004/02/skos/core#notation> "10002198";
<http://www.w3.org/2004/02/skos/core#prefLabel> "Anaphylactic reaction" .
<http://purl.bioontology.org/ontology/MDR/10002199> a <http://www.w3.org/2004/02/skos/core#Concept>;
<http://purl.bioontology.org/ontology/MDR/level> "PT";
<http://www.w3.org/2004/02/skos/core#broader> <http://purl.bioontology.org/ontology/MDR/10002220>,
<http://purl.bioontology.org/ontology/MDR/10009193>;
<http://www.w3.org/2004/02/skos/core#inScheme> <http://purl.bioontology.org/ontology/MDR>;
<http://www.w3.org/2004/02/skos/core#notation> "10002199";
<http://www.w3.org/2004/02/skos/core#prefLabel> "Anaphylactic shock" .
<http://purl.bioontology.org/ontology/MDR/10020751> a <http://www.w3.org/2004/02/skos/core#Concept>;
<http://purl.bioontology.org/ontology/MDR/level> "PT";
<http://www.w3.org/2004/02/skos/core#broader> <http://purl.bioontology.org/ontology/MDR/10027654>;
<http://www.w3.org/2004/02/skos/core#inScheme> <http://purl.bioontology.org/ontology/MDR>;
<http://www.w3.org/2004/02/skos/core#notation> "10020751";
<http://www.w3.org/2004/02/skos/core#prefLabel> "Hypersensitivity" .
<http://purl.bioontology.org/ontology/MDR/10067484> a <http://www.w3.org/2004/02/skos/core#Concept>;
<http://purl.bioontology.org/ontology/MDR/level> "PT";
<http://www.w3.org/2004/02/skos/core#broader> <http://purl.bioontology.org/ontology/MDR/10043409>;
<http://www.w3.org/2004/02/skos/core#inScheme> <http://purl.bioontology.org/ontology/MDR>;
<http://www.w3.org/2004/02/skos/core#notation> "10067484";
<http://www.w3.org/2004/02/skos/core#prefLabel> "Adverse reaction" .

38. Mapping Provenance Representation
38
:NanoPub_1_Supporting_2 = {
[ a r:Proof, r:Conjunction;
r:component <#lemma1>;
r:component <#lemma2>;
r:component <#lemma3>;
r:component <#lemma4>;
r:component <#lemma5>;
r:component <#lemma6>;
r:gives {
<http://purl.bioontology.org/ontology/ICD9CM/999.4> skos:prefLabel "Anaphylactic shock due to serum, not elsewhere classified".
<http://purl.bioontology.org/ontology/SNOMEDCT/213320003> skos:prefLabel "Anaphylactic shock due to serum".
<http://purl.bioontology.org/ontology/MDR/10067113> skos:prefLabel "Anaphylactic transfusion reaction".
<http://purl.bioontology.org/ontology/ICD9CM/999.4> skos:exactMatch <http://purl.bioontology.org/ontology/SNOMEDCT/213320003>.
<http://purl.bioontology.org/ontology/SNOMEDCT/213320003> skos:exactMatch <http://purl.bioontology.org/ontology/MDR/10067113>.
<http://purl.bioontology.org/ontology/ICD9CM/999.4> skos:exactMatch <http://purl.bioontology.org/ontology/MDR/10067113>.
}].
…….
…….
…….
<#lemma13> a r:Inference; r:gives {<http://purl.bioontology.org/ontology/ICD9CM/999.4> skos:exactMatch
<http://purl.bioontology.org/ontology/MDR/10067113>}; r:evidence (
<#lemma11>
<#lemma12>);
r:binding [ r:variable [ n3:uri "http://localhost/var#x0"]; r:boundTo [ n3:uri "http://purl.bioontology.org/ontology/ICD9CM/999.4"]];
r:binding [ r:variable [ n3:uri "http://localhost/var#x1"]; r:boundTo [ n3:uri "http://purl.bioontology.org/ontology/SNOMEDCT/213320003"]];
r:binding [ r:variable [ n3:uri "http://localhost/var#x2"]; r:boundTo [ n3:uri "http://purl.bioontology.org/ontology/MDR/10067113"]];
r:rule <#lemma14>.
<#lemma14> a r:Extraction; r:gives {@forAll var:x0, var:x1, var:x2. {var:x0 skos:exactMatch var:x1.
var:x1 skos:exactMatch var:x2} => {var:x0 skos:exactMatch var:x2}};
r:because [ a r:Parsing; r:source <file:///Users/sajjad/workspace/terminology-reasoning-test-case/example-term-map.n3>].
}.