Schemas and Ontologies: Building a Semantic Infrastructure for the GRID and Digital Libraries Speakers Abstract & Bio

[Return to Programme] [Return to Event Page] [Return to eSI Programme List]

Jeremy Rogers qualified from Manchester medical school in 1989 and attained membership of the Royal College of General Practitioners in 1993. He joined the Medical Informatics Group at the University of Manchester in 1994 as a full time Clinical Research Fellow, although still maintains a low level of clinical practice.

His primary research focus is the development, population and exploitation of medical ontologies such as those underlying the OpenGALEN technologies and SNOMED-CT. Wherever possible the research has been practically grounded in projects to develop end-user clinical tools, particularly clinical data entry or decision support tools, such as the recent PRODIGY and Prescribing Indicators projects.

Current projects include the MRC eScience CLEF initiative, which aims to construct a repository of cancer patient records indexed by their principal diagnoses, treatments and comorbidities. A key part of the project is to marry text extraction and ontology technologies in order that the necessary information can be extracted from the body of individual patient records.

Jeremy has attended and presented papers at a number of international medical and bio-informatics conferences, including AMIA, MEDINFO, MIE, TEPR and TEHRE. He has provided tutoring and course content for the Royal College of Surgeons of Edinburgh distance learning Diploma in Medical Informatics, and consulted for Mayo Clinic.

Why Ontologies ?

Humans have a great fondness for categorising the world they observe around them. Most scientists are familiar with the problems that arise when quantitative principles of division are applied: accuracy at initial data capture is paramount, whilst techniques for analysing numerical data are relatively well understood. In the medical and bioscience domains, however, many of the properties used for differentiating or aggregating objects are descriptive and not quantifiable: what biological function do these proteins perform ? What kinds of cancers produce this protein ?

The talk will focus on the medical domain, and its long tradition of attempting to organise and classify its descriptive terms. As the number of medical terms available has increased, and as different users have demanded different principles for their classification, the speaker will show how the size and complexity of the task has outstripped the ability of unaided humans to provide coherent or useable structures. The speaker will propose (but with caveats) that ontologies and ontological engineering tools and methodologies provide necessary aid to this and similar tasks.