gms | German Medical Science

Introduction and background: The precise definition and semantically correct representation of documentation features and study items has an important impact on the value and quality of clinical trials. It is an essential assumption to support the reuse of such features, to compare the data of different study projects, and to establish a common basis for the interpretation of the resulting data. We report on work in progress of a long term project, initiated in [1], which aims at the development of an ontology of study items and of clinical phenotypes. There is a close relation between the notions of study item, data element, property, and clinical phenotype.

Methods: The usage of the General Formal Ontology (GFO) [2] is a part of the method, whose basic idea consists in the ontological reconstruction and modelling of entities of the medical domain [3]. This reconstruction uses the ontology of properties, as expounded in GFO, as well as some basic constructs of the ISO-11179-standard, notably the notion of data element [4].

Results: We propose a definition of study items which are described as particular data elements in ISO-11179 [1], whose conceptual structure is ontologically reconstructed and uncovered within the GFO. Furthermore, we propose a definition of the notion of a clinical phenotype and relate it to the notion of data element and to the notion of property in GFO. We outline a formal ontological framework, based on GFO, to treat clinical phenotypes, properties and study items in a uniform manner. This framework takes up the approach in [5] and makes a further step in establishing a conceptual tool for the specification of clinical phenotypes. Furthermore, we implemented a first version of a prototype for managing study items. Study items must be captured and specified in a semantically correct way and computationally presented such that they can be efficiently retrieved and used.

Discussion and conclusion: The final aim of the presented research is the development of ontologies of clinical phenotpyes. To achieve such ontologies, a number of ontological relations must be introduced and specified, which connect phenotypes to form complex properties. These relations must include, among others, part-of, participation in a process, causality, function realization. Furthermore, a suitable process ontology must be integrated to specify physiological information.