BIOMEDICAL ONTOLOGIES: EXAMINING ASPECTS OF INTEGRATION ACROSS BREAST CANCER KNOWLEDGE DOMAINS

The key ideas developed in this thesis lie at the intersection of epistemology, philosophy of molecular biology, medicine, and computer science. I examine how the epistemic and pragmatic needs of agents distributed across particular scientific disciplines influence the domain-specific reasoning, classification, and representation of breast cancer. The motivation to undertake an interdisciplinary approach, while addressing the problems of knowledge integration, originates in the peculiarity of the integrative endeavour of sciences that is fostered by information technologies and ontology engineering methods. I analyse what knowledge integration in this new field means and how it is possible to integrate diverse knowledge domains, such as clinical and molecular. I examine the extent and character of the integration achieved through the application of biomedical ontologies. While particular disciplines target certain aspects of breast cancer-related phenomena, biomedical ontologies target biomedical knowledge about phenomena that is often captured within diverse classificatory systems and domain-specific representations. In order to integrate dispersed pieces of knowledge, which is distributed across assorted research domains and knowledgebases, ontology engineers need to deal with the heterogeneity of terminological, conceptual, and practical aims that are not always shared among the domains. Accordingly, I analyse the specificities, similarities, and diversities across the clinical and biomedical domain conceptualisations and classifications of breast cancer. Instead of favouring a unifying approach to knowledge integration, my analysis shows that heterogeneous classifications and representations originate from different epistemic and pragmatic needs, each of which brings a fruitful insight into the problem. Thus, while embracing a pluralistic view on the ontologies that are capturing various aspects of knowledge, I argue that the resulting integration should be understood in terms of a coordinated social effort to bring knowledge together as needed and when needed, rather than in terms of a unity that represents domain-specific knowledge in a uniform manner. Furthermore, I characterise biomedical ontologies and knowledgebases as a novel socio-technological medium that allows representational interoperability across the domains. As an example, which also marks my own contribution to the collaborative efforts, I present an ontology for HER2+ breast cancer phenotypes that integrates clinical and molecular knowledge in an explicit way. Through this and a number of other examples, I specify how biomedical ontologies support a mutual enrichment of knowledge across the domains, thereby enabling the application of molecular knowledge into the clinics.