Machine learning with biomedical ontologies

Kulmanov, Maxat; Smaili, Fatima Zohra; Gao, Xin; Hoehndorf, Robert

doi:10.1101/2020.05.07.082164

Cited by 24 publications

(18 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our previous work has also demonstrated that expansion of vocabulary [25] and ontology extension [26] can improve performance of a similar tasks. Recent work has also explored alternative methods for employing ontology axioms and taxonomy for classification and ranking problems, such as the conversion of ontology axioms to vectors [27], an approach which has been demonstrated to improve performance when compared semantic similarity approaches [28].…”

Section: Resultsmentioning

confidence: 99%

“…Our previous work also showed that extension of ontologies by examining binary relations mined from text, and extension of ontology vocabularies with information from other ontologies, improved performance at a semantic similarity-based patient characterisation tasks [26, 27]. Recent work has also explored alternative methods for employing ontology axioms and taxonomy for classification and ranking problems, such as the conversion of ontology axioms to vectors [28], an approach which has been demonstrated to improve performance when compared semantic similarity approaches [29].…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Towards Similarity-based Differential Diagnostics For Common Diseases

Slater

Karwath

Williams

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Ontology-based phenotype profiles have been utilised for the purpose of differential diagnosis of rare genetic diseases, and for decision support in specific disease domains. Particularly, semantic similarity facilitates diagnostic hypothesis generation through comparison with disease phenotype profiles. However, the approach has not been applied for differential diagnosis of common diseases, or generalised clinical diagnostics from uncurated text-derived phenotypes. In this work, we describe the development of an approach for deriving patient phenotype profiles from clinical narrative text, and apply this to text associated with MIMIC-III patient visits. We then explore the use of semantic similarity with those text-derived phenotypes to classify primary patient diagnosis, comparing the use of patient-patient similarity and patient-disease similarity using phenotype-disease profiles previously mined from literature. The results reveal a potentially powerful approach that can be applied to a variety of clinical tasks, such as differential diagnosis, cohort discovery, document and text classification, and outcome prediction.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Towards Similarity-based Differential Diagnostics For Common Diseases

Slater

Karwath

Williams

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…To provide DL2Vec with structured background knowledge of human and mouse phenotypes as well as protein functions, we used the crossspecies phenotype ontology PhenomeNET (Hoehndorf et al, 2011;Rodríguez-García et al, 2017), which is built upon and includes the Gene Ontology (Ashburner et al, 2000; The Gene Ontology Consortium, 2017). These ontologies contain formalized biological background knowledge (Hoehndorf et al, 2015b), which has the potential to significantly improve the performance of these features in machine learning and predictive analyses (Smaili et al, 2019;Kulmanov et al, 2020). DeepViral consists of a phenotype model trained on phenotypes caused by a viral infection and a sequence model trained on protein sequences, as shown in Figure 1 (b).…”

Section: Embedding Features Of Viruses and Human Proteins From Phenotmentioning

confidence: 99%

DeepViral: infectious disease phenotypes improve prediction of novel virus–host interactions

Liu-Wei

Kafkas

Chen

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Motivation: Infectious diseases from novel viruses are becoming a major public health concern. Fast identification of virus-host interactions can reveal mechanistic insights of infectious diseases and shed light on potential treatments and drug discoveries. Current computational prediction methods for novel viruses are based only on protein sequences. Yet, it is not clear to what extent other important features, such as the symptoms caused by the viruses, could contribute to a predictor. Disease phenotypes (i.e., symptoms) are readily accessible from clinical diagnosis and we hypothesize that they may act as a potential proxy and an additional source of information for the underlying molecular interactions between the pathogens and hosts. Results: We developed DeepViral, a deep learning method that predicts potential protein-protein interactions between human and viruses. First, human proteins and viruses were embedded in a shared space using their associated phenotypes, functions, taxonomic classification, as well as formalized background knowledge from biomedical ontologies. By extending a sequence learning model with phenotype features, our model can not only significantly improve over previous sequence-based approaches for inter-species interaction prediction, but also identify pathways of viral targets under a realistic experimental setup for novel viruses. Availability:https://github.com/bio-ontology-research-group/DeepViral Contact:

show abstract

“…They also provide a rich semantic underpinning by means of description logics, enabling semantic information retrieval and analysis. Recent works in ontology-based machine learning and vectorisation, such as OPA2Vec [4], are bridging the gap between text mining, machine learning, and semantic analysis, and they highlight the need for novel approaches that make full use of the descriptive and semantic features provided by biomedical ontologies [5].…”

Section: Introductionmentioning

confidence: 99%

Komenti: A semantic text mining framework

Slater

Bradlow

Hoehndorf

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Komenti is a reasoner-enabled semantic query and information extraction tool. It is the only text mining tool that enables querying inferred knowledge from biomedical ontologies. It also contains multiple novel components for vocabulary construction and context disambiguation, which can improve the power of text mining and ontology-based analysis tasks, with a view towards making full use of the semantic provision of biomedical ontologies in the text extraction and characterisation space. Here, we describe Komenti, its features, and a use case wherein we automate a clinical audit process, classifying the medications of patients with hypertrophic cardiomyopathy from text records, revealing a high precision, and a subcohort of candidate patients who have atrial fibrillation but were not anti-coagulated, and are therefore at a higher risk of stroke.

show abstract

Machine learning with biomedical ontologies

Cited by 24 publications

References 24 publications

Towards Similarity-based Differential Diagnostics For Common Diseases

Towards Similarity-based Differential Diagnostics For Common Diseases

DeepViral: infectious disease phenotypes improve prediction of novel virus–host interactions

Komenti: A semantic text mining framework

Contact Info

Product

Resources

About