Improved characterisation of clinical text through ontology-based vocabulary expansion

Background Biomedical ontologies contain a wealth of metadata that constitutes a fundamental infrastructural resource for text mining. For several reasons, redundancies exist in the ontology ecosystem, which lead to the same entities being described by several concepts in the same or similar contexts across several ontologies. While these concepts describe the same entities, they contain different sets of complementary metadata. Linking these definitions to make use of their combined metadata could lead to improved performance in ontology-based information retrieval, extraction, and analysis tasks. Results We develop and present an algorithm that expands the set of labels associated with an ontology class using a combination of strict lexical matching and cross-ontology reasoner-enabled equivalency queries. Across all disease terms in the Disease Ontology, the approach found 51,362 additional labels, more than tripling the number defined by the ontology itself. Manual validation by a clinical expert on a random sampling of expanded synonyms over the Human Phenotype Ontology yielded a precision of 0.912. Furthermore, we found that annotating patient visits in MIMIC-III with an extended set of Disease Ontology labels led to semantic similarity score derived from those labels being a significantly better predictor of matching first diagnosis, with a mean average precision of 0.88 for the unexpanded set of annotations, and 0.913 for the expanded set. Conclusions Inter-ontology synonym expansion can lead to a vast increase in the scale of vocabulary available for text mining applications. While the accuracy of the extended vocabulary is not perfect, it nevertheless led to a significantly improved ontology-based characterisation of patients from text in one setting. Furthermore, where run-on error is not acceptable, the technique can be used to provide candidate synonyms which can be checked by a domain expert.

Improved characterisation of clinical text through ontology-based vocabulary expansion

BackgroundBiomedical ontologies contain a wealth of metadata that constitutes a fundamental infrastructural resource for text mining. For several reasons, redundancies exist in the ontology ecosystem, which lead to the same concepts being described by several terms in the same or similar contexts across several ontologies. While these terms describe the same concepts, they contain different sets of complementary metadata. Linking these definitions to make use of their combined metadata could lead to improved performance in ontology-based information retrieval, extraction, and analysis tasks.ResultsWe develop and present an algorithm that expands the set of labels associated with an ontology class using a combination of strict lexical matching and cross-ontology reasoner-enabled equivalency queries. Across all disease terms in the Disease Ontology, the approach found 51,362 additional labels, more than tripling the number defined by the ontology itself. Manual validation by a clinical expert on a random sampling of expanded synonyms over the Human Phenotype Ontology yielded a precision of 0.912. Furthermore, we found that annotating patient visits in MIMIC-III with an extended set of Disease Ontology labels led to semantic similarity score derived from those labels being a significantly better predictor of matching first diagnosis, with a mean average precision of 0.88 for the unexpanded set of annotations, and 0.913 for the expanded set.ConclusionsInter-ontology synonym expansion can lead to a vast increase in the scale of vocabulary available for text mining applications. While the accuracy of the extended vocabulary is not perfect, it nevertheless led to a significantly improved ontology-based characterisation of patients from text in one setting. Furthermore, where run-on error is not acceptable, the technique can be used to provide candidate synonyms which can be checked by a domain expert.

Using Semantics for morphological Descriptions in Morph•D•Base

Providing data in a semantically structured format has become the gold standard in data science. However, a significant amount of data is still provided as unstructured text - either because it is legacy data or because adequate tools for storing and disseminating data in a semantically structured format are still missing. We have developed a description module for Morph∙D∙Base, a semantic knowledge base for taxonomic and morphologic data, that enables users to generate highly standardized and formalized descriptions of anatomical entities using free text and ontology-based descriptions. The main organizational backbone of a description in Morph∙D∙Base is a partonomy, to which the user adds all the anatomical entities of the specimen that they want to describe. Each element of this partonomy is an instance of an ontology class and can be further described in two different ways: as semantically enriched free-text description that is annotated with terms from ontologies, and semantically through defined input forms with a wide range of ontology-terms to choose from. To facilitate the integration of the free text into a semantic context, text can be automatically annotated using jAnnotator, a javascript library that uses about 700 ontologies with more than 8.5 million classes of the National Center for Biomedical Ontology (NCBO) bioportal. Users get to choose from suggested class definitions and link them to terms in the text, resulting in a semantic markup of the text. This markup may also include labels of elements that the user already added to the partonomy. Anatomical entities marked in the text can be added to the partonomy as new elements that can subsequently be described semantically using the input forms. Each free text together with its semantic annotations is stored following the W3C Web Annotation Data Model standard (https://www.w3.org/TR/annotation-model). The whole description with the annotated free text and the formalized semantic descriptions for each element of the partonomy are saved in the tuplestore of Morph∙D∙Base. The demonstration is targeted at developers and users of data portals and will give an insight to the semantic Morph∙D∙Base knowledge base (https://proto.morphdbase.de) and jAnnotator (http://git.morphdbase.de/christian/jAnnotator).