What's a "disease"? Questions for applied ontologies of diseases

This issue of Applied Ontology includes an article by Rovetto and Mizoguchi (2015) that sets out an alternative ontological formulation of the notion of “diseases” as “causal chains” or “processes”, which they call the “River Flow Model” (RFM). They contrast their account with that in the Ontology of General Medical Science (OGMS) as described by Scheuermann et al. (2009) that formalises diseases as “dispositions” to undergo “pathological processes” as the result of “disorders” that are kinds of “physical components”. There have also been a series of articles published recently by Schulz and his colleagues arguing that the entities in the large coding system for diseases, SNOMED CT (IHTSDO, 2015), are best interpreted as “situations” (Schulz et al., 2011) by which they define as “phases of the life of a patient, during which he/she is bearer of a clinical condition”. This interpretation has been largely accepted by both the SNOMED organisation and the group harmonising the latest revision of the World Health Organisation’s International Classification of Diseases (ICD) with SNOMED CT (Schulz et al., 2012). In addition, entities entitled “disease” occur in at least 45 of the ontologies in BioPortal (NCBO, 2015), as well as in other standards such as the de facto standard for exchange of laboratory test data, LOINC (LOINC, 2015), the National Library of Medicine’s Unified Medical Language System’s Semantic Network (UMLS, 2003), and various more general resources such as WordNet (WordNet, 2015). Some of these are not “ontologies” in the sense usually understood in this Journal. The word “ontology” is now used for such a variety of information artefacts that, without qualification, it is almost meaningless. We take it here in a relatively narrow sense familiar to most readers of this Journal as a formal representation of the definitions and necessary characteristics of the entities in the world, or more broadly, of the entities that appear in some information system. Given this situation, how should we discuss and evaluate these various ontological formulations of “diagnosis” for purposes of applications? We take it as given that the purpose of the ontological. Although Rovetto and Mizoguchi discuss their application briefly, as with most papers in this area the focus is primarily on the internal structure consistency of the ontology. So, how should a developer decide which