Commentary on “Epilepsy informatics and an ontology-driven infrastructure for large database research and patient care in epilepsy”
Article first published online: 30 JUL 2013
Copyright © 2013 International League Against Epilepsy
Volume 54, Issue 8, pages 1507–1509, August 2013
How to Cite
Bergin, P. (2013), Commentary on “Epilepsy informatics and an ontology-driven infrastructure for large database research and patient care in epilepsy”. Epilepsia, 54: 1507–1509. doi: 10.1111/epi.12239
- Issue published online: 30 JUL 2013
- Article first published online: 30 JUL 2013
When I was in training, a senior neurologist passed on an aphorism that he had been taught when he himself was a junior: “You don't understand neurology until you understand epilepsy, and you don't understand epilepsy until you have reclassified it.” I am not sure where this quip originated, but it makes an important point: epilepsy is difficult. The difficulty in understanding epilepsy reflects the extraordinary complexity of the human brain. As a result of this complexity, there are many different types of seizures and many reasons that people develop epilepsy.
In their recent report, the International League Against Epilepsy (ILAE) commission on classification and terminology repeatedly emphasize how limited our current understanding of epilepsy is, and the subsequent difficulties in developing a comprehensive classification (Berg et al., 2010). They suggest that their own proposals regarding classification should be seen as a work in progress; they acknowledge that the classification system they proposed is not a definitive one, and will certainly change as our understanding of pathogenesis of seizures improves.
In their article, Sahoo et al. discuss some of the issues raised by the ILAE commission (Sahoo et al., 2013). Many of these relate to the fundamental complexities of epilepsy and our inadequate understanding of the causes. In addition, epilepsy involves a diverse stakeholder community. Sahoo et al. emphasize that all people who use particular terms should mean the same things by them. They discuss problems that result from both semantic heterogeneity (when identical terms are used to describe heterogeneous information) and syntactic heterogeneity (differences in the format of data representation).
These are issues that have also been taxing those of working on the EpiNet project (Bergin et al., 2012). The EpiNet project was established to facilitate investigator-initiated clinical research in epilepsy. It has been designed to conduct multicenter cohort studies and to run simple pragmatic clinical trials. How do we ensure that all those who enter patients into the EpiNet database actually mean the same thing when they use specific terms? Does an epileptologist in one country necessarily interpret terms in the same way as a neurologist in a different country, or a different continent?
Clearly, these are important issues for a project such as EpiNet, which welcomes participation from a wide range of neurologists and epileptologists. If we cannot be sure that people mean the same thing by specific terms, then people will not have any confidence in the results that are produced. It is critical that the “garbage in, garbage out” scenario is prevented.
EpiNet will address some of these issues by undertaking a series of validation studies. However, it would certainly be of major benefit to us—and presumably others involved in research—if there was widespread acceptance of meanings for specific terms, and how different conditions relate to one another.
Sahoo et al. propose a solution to many of these issues. They suggest that the epilepsy community undertakes a formal classification process using “informatics.” Their proposal is that the epilepsy community develops an “ontology.” The ontology should incorporate well-defined semantics that allow precise definition. The ontology would comprise a “formal representation of knowledge in the domain of epilepsy that allows both human users and machines to consistently and accurately interpret terms.” Epilepsy terms would be classified along a number of distinct dimensions according to specific application requirements. Named relationships would be used to describe the various subcomponents of structures, as well as functional relationships. Sahoo et al. point out that introducing an ontology would have multiple benefits, including improving clinical decision support, access to information for patients and their families, data sharing, and secondary use of clinical data. Another benefit would be that an epilepsy ontology would be able to interact with other ontologies outside of epilepsy. Sahoo et al. state that the integrated terminology system associated with the epilepsy ontology will make software applications more reliable, because the ontology allows accurate interpretation of the terms. However, to do this, Sahoo et al. assert that it is important to “move beyond a document-based encoding to use of a formal knowledge representation language that can be directly integrated into informatics tools.” They discuss other problems involved with creating an ontology, and make the particular comment that “ontology languages have not so far had a straightforward mechanism to address the issue of dynamic classification of terms.”
As someone involved with clinical research, I was very interested by the proposal of Sahoo et al., since it addresses several of the issues with which we at the EpiNet project have struggled. I had relatively little difficulty understanding the broad concepts. However, I did have difficulty with some of the detail. The authors use terms that are presumably familiar to those working in the field of informatics, but which are not necessarily in common use by those working in the field of epilepsy; these include terms such as granularity, interoperability, and even informatics itself. They talk of “specialization-generalization relationships,” “ two-dimensional matrix,” and “formal knowledge representation language,” although the terms are not well defined.
In the longer term, the detail will be important, but at this stage I think Sahoo et al. are seeking agreement on general principles. To this end, I agree that the goal is certainly one worth striving for. If the epilepsy community can successfully create an ontology that is universally accepted, then it will be a major advance, particularly with regard to research, but also in terms of management of individual patients. However, it will not be easy. I fear that Sahoo et al. may have underestimated the difficulty in producing the epilepsy ontology. I genuinely hope that the epilepsy community can reach a consensus in this area, but Sahoo et al. themselves noted the controversy that has followed the attempt by the commission on classification and terminology to move in this area. Notwithstanding these concerns, I think the proposal to create an epilepsy ontology is an excellent one, and I remain cautiously optimistic that it will eventually be successful. I am encouraged to learn that there has been widespread acceptance of the ontologies created in the fields of genetics and oncology. Those of us working on the EpiNet project would be happy to participate in the development of an epilepsy ontology.
I have no conflicts of interest to disclose. I confirm that I have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
- 2010) Revised terminology and concepts for organization of seizures and epilepsies: report of the ILAE Commission on Classification and Terminology, 2005–2009. Epilepsia 51:676–685. , , , , , , , , , , , , , . (
- EpiNet study group. (2012) An International Pilot Study of an Internet-Based Platform to facilitate clinical research in epilepsy; the EpiNet Project. Epilepsia, 53:1829–1835. , , , , , , , , , , , , , , , , ;
- 2013) Epilepsy informatics and an ontology-driven infrastructure for large database research and patient care in epilepsy. Epilepsia May 3 [Epub ahead of print]. , , . (