The usefulness of ontologies in health
research: focus on epidemiology
Claudia Pagliari PhD FRCPE
eHealth Research Group
The University of Edinburgh
Ontology Network meeting, Edinburgh April
In which areas of medicine are
ontologies used?
• Management & exchange of clinical data
(drugs, symptoms, history, patient characteristics
• Medical knowledge collation for evidencebased practice
• Data reuse for epidemiological research
• Information-intensive biomedical research
Bowman (2005). "Coordinating SNOMED-CT and ICD-10: Getting the Most out of Electronic
Health Record Systems." Journal of AHIMA 76, no.7 , 60-61.
Chan, J. and Kishore, R. and Sternberg, P. and Van Auken, K. (2012) The Gene Ontology:
enhancements for 2011. Nucleic Acids Research, 40 (D1). D559-D564. ISSN 0305-1048
[Cradle-to-grave Patient Data in NHS Scotland]
3 phases of ontologies in eHealth
1) Focus on terminology & coding schemes. Dominated by library
science. Oriented towards cataloguing & indexing of published
literature – e.g. MESH
2) Focus on database design & software. Dominated by
programmers. Oriented towards describing & promoting access
to data – e.g. HL7 Reference Information Model
3) Focus on biological and patient reality. Biologists more
involved in ontology development – e.g. Gene Ontology (GO);
Open Biomedical Ontologies (OBO)
Smith & Brochhausen (2012), “Establishing and Harmonizing Ontologies in an Interdisciplinary Health Care and Clinical Research
Environment”, in: B. Blobel P. Pharow and M. Nerlich (eds.), eHealth: Combining Health Telematics, Telemedicine, Biomedical
Engineering and Bioinformatics on the Edge (Global Expert Summit Textbook, Studies in Health, Technology and Informatics, 134),
IOS Press, Amsterdam, 219-234
Data linkage in epidemiology &
the need for ontologies
Data linkage initiatives spawning…
Population-based observational
– Epidemiological studies (frequencies, trends,
– Natural experiments – (evaluation of new
interventions or policies e.g. smoking ban)
– Pharmacosurveillance (e.g. WoSCoPS)
Ontologies & semantics in
• Epidemiology research is multidisciplinary
• Requires integrative frameworks for data sharing
• Semantic Web may help to:
improve interoperability
enable machines to trace & process resources
converge epidemiological & predictive models
find hidden patterns in epidemiological data
• Requires effective use of ontologies (rate-limiting)
• Need to evaluate ontologies against outcomes &
prioritise quality
Ferreira (2013) On the usefulness of ontologies in epidemiological research. J Epidemiol Community Health, Vol 67 No 5
Hohendorf (2012) Evaluation of research in biomedical ontologies. Briefings in Bioinformatics Advance Access, Sept 8

Slides - University of Edinburgh