RESUMO
BACKGROUND: Biosample collections and biobank information systems have become a key enabler for medical research. Therefore it is important to identify potentially relevant ontologies to semantically enrich information related to the biobanking domain. OBJECTIVES: We present a three-stage semi-automated evaluation approach which allows identifying relevant ontologies for the biobanking domain based on competency questions. METHODS: After identifying candidate biobanking ontologies (Stage 1) and competency questions (Stage 2), a six-step lexical evaluation approach, which assesses the coverage of concepts, properties or instances defined by competency questions is suggested and described (Stage 3). RESULTS: We were able to perform a proof-of-concept evaluation of the OMIABIS ontology using our proposed three-stage approach together with a sample competency question. CONCLUSION: Our evaluation approach allows a swift evaluation of candidate ontology entities based on a search for higher hierarchy key terms that exist in comprehensive medical vocabularies in order to state the usability of specific ontologies for the biobanking domain.
Assuntos
Ontologias Biológicas , Bancos de Espécimes Biológicos/classificação , Uso Significativo , Processamento de Linguagem Natural , Terminologia como Assunto , Áustria , Aprendizado de MáquinaRESUMO
The knowledge about the quality of samples and associated clinical data in biospecimen collections is a premise of clinical research. An electronic biosample register aims to facilitate the discovery of information about biosample collections in a hospital. Moreover, it might improve scientific collaboration and research quality through a shared access to harmonized sample collection description data. The aim of this paper is to present a concept of a web-based biosample register of the existing biosample collections at the Medical University of Innsbruck. A uniform description model is built based on an analysis of the sample collection data of independent sample management systems from two departments within the hospital. An extended set of attributes of the minimum dataset used by the Swedish sample collection register (MIABIS) has been applied to all biosample collections as a common description model. The results of the analysis and the data model are presented together with a first concept of a sample collection search register.