Towards applying text mining and natural language processing for biomedical ontology acquisition

Inniss, Tasha R.; Lee, John Y.K.; Light, Marc; Grassi, Michael A.; Thomas, George; Williams, Andrew B.

doi:10.1145/1183535.1183539

Cited by 12 publications

(6 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Collocations in a corpus of clinical notes are prime candidates to be mapped to meaningful phenotypes [19-21]. Collocations can also help uncover multi-word terms that are not covered by medical terminologies.…”

Section: Introductionmentioning

confidence: 99%

Redundancy in electronic health record corpora: analysis, impact on text mining performance and mitigation strategies

Cohen

Elhadad

2013

BMC Bioinformatics

View full text Add to dashboard Cite

BackgroundThe increasing availability of Electronic Health Record (EHR) data and specifically free-text patient notes presents opportunities for phenotype extraction. Text-mining methods in particular can help disease modeling by mapping named-entities mentions to terminologies and clustering semantically related terms. EHR corpora, however, exhibit specific statistical and linguistic characteristics when compared with corpora in the biomedical literature domain. We focus on copy-and-paste redundancy: clinicians typically copy and paste information from previous notes when documenting a current patient encounter. Thus, within a longitudinal patient record, one expects to observe heavy redundancy. In this paper, we ask three research questions: (i) How can redundancy be quantified in large-scale text corpora? (ii) Conventional wisdom is that larger corpora yield better results in text mining. But how does the observed EHR redundancy affect text mining? Does such redundancy introduce a bias that distorts learned models? Or does the redundancy introduce benefits by highlighting stable and important subsets of the corpus? (iii) How can one mitigate the impact of redundancy on text mining?ResultsWe analyze a large-scale EHR corpus and quantify redundancy both in terms of word and semantic concept repetition. We observe redundancy levels of about 30% and non-standard distribution of both words and concepts. We measure the impact of redundancy on two standard text-mining applications: collocation identification and topic modeling. We compare the results of these methods on synthetic data with controlled levels of redundancy and observe significant performance variation. Finally, we compare two mitigation strategies to avoid redundancy-induced bias: (i) a baseline strategy, keeping only the last note for each patient in the corpus; (ii) removing redundant notes with an efficient fingerprinting-based algorithm. aFor text mining, preprocessing the EHR corpus with fingerprinting yields significantly better results.ConclusionsBefore applying text-mining techniques, one must pay careful attention to the structure of the analyzed corpora. While the importance of data cleaning has been known for low-level text characteristics (e.g., encoding and spelling), high-level and difficult-to-quantify corpus characteristics, such as naturally occurring redundancy, can also hurt text mining. Fingerprinting enables text-mining techniques to leverage available data in the EHR corpus, while avoiding the bias introduced by redundancy.

show abstract

Section: Introductionmentioning

confidence: 99%

Redundancy in electronic health record corpora: analysis, impact on text mining performance and mitigation strategies

Cohen

Elhadad

2013

BMC Bioinformatics

View full text Add to dashboard Cite

show abstract

“…Inniss et al (2006) utilizam técnicas de mineração de textos e PLN para extrair, a partir de descrições verbais transcritas de especialistas em retina, pares características-atributos de degeneração macular relacionada à idade. Esses pares são utilizados para produzir colaborativamente uma ontologia nesse domínio.…”

Section: Desenvolvimento De Ontologiasunclassified

Mineração de textos biomédicos: uma revisão bibliométrica

Woszezenki¹,

Gonçalves

2013

Perspect. ciênc. inf.

View full text Add to dashboard Cite

A mineração de textos vem sendo, cada vez mais, empregada para automatizar o processo de extração de informações importantes, contidas em textos biomédicos, possibilitando que os pesquisadores fiquem a par do desenvolvimento da biomedicina. Considerando a importância deste campo de pesquisa, este artigo apresenta um mapeamento das publicações científicas sobre mineração de textos biomédicos e discute as principais tarefas desse campo de pesquisa, as quais os pesquisadores têm dedicado maior atenção. Para isso, foi utilizada a bibliometria, uma técnica que permite analisar o desenvolvimento de um campo da ciência, visando identificar suas características. O mapeamento apresentado promove o conhecimento sobre o histórico e o estado atual do campo de pesquisa e disponibiliza insumos, que permitem enriquecer a discussão sobre os possíveis rumos que as pesquisas, na área, têm tomado e as prováveis tendências científicas para os pesquisadores e interessados no tema.Palavras-chave: Mineração de textos; Descoberta baseada em literatura; Bioinformática.

show abstract

Section: Introductionmentioning

confidence: 99%

“…A lot of this information is stored as narrative texts, hindering its use for the decision making processes. The process of discovering the knowledge contained in these texts, or creating new hypotheses according to them include a lot of time and effort [14], [24] that cannot be afforded by most organizations.Generally, this problem remains a constant challenge due to the difficulty of organizations to absorb and use the information they need. It should be noted that, the limits on the reading speed of a human being make it impossible to capture the key information in a short time when there is a large amount of text [18].…”

mentioning

confidence: 99%

Great Model: A Model for the Automatic Generation of Semantic Relations Between Text Summaries Based on the Relevant Information to the User

Gomez¹,

Pomares²

2013

IJCSITY

View full text Add to dashboard Cite

Abstract.The large available amount of non-structured texts that be-long to different domains such as healthcare (e.g. medical records), justice (e.g. laws, declarations), insurance (e.g. declarations), etc. increases the effort required for the analysis of information in a decision making pro-cess. Different projects and tools have proposed strategies to reduce this complexity by classifying, summarizing or annotating the texts. Partic-ularly, text summary strategies have proven to be very useful to provide a compact view of an original text. However, the available strategies to generate these summaries do not fit very well within the domains that require take into consideration the temporal dimension of the text (e.g. a recent piece of text in a medical record is more important than a pre-vious one) and the profile of the person who requires the summary (e.g the medical specialization). To cope with these limitations this paper presents "GReAT" a model for automatic summary generation that re-lies on natural language processing and text mining techniques to extract the most relevant information from narrative texts and discover new in-formation from the detection of related information. GReAT Model was implemented on software to be validated in a health institution where it has shown to be very useful to display a preview of the information about medical health records and discover new facts and hypotheses within the information. Several tests were executed such as Functional-ity, Usability and Performance regarding to the implemented software. In addition, precision and recall measures were applied on the results ob-tained through the implemented tool, as well as on the loss of information obtained by providing a text more shorter than the original. Keywords:Topic Identification, User's Profile, Text Mining, Automatic Summary Generation Techniques, Semantic Relations This work was supported by the project "Extraccion semi-autom´atica de metadatos de fuentes de datos estructuradas: Una aproximacion basada en agentes y miner´ıa de datos" funded by Banco Santander S.A. and Pontificia Universidad Javeriana. 1.IntroductionDuring the last thirty years the information systems have stored huge amounts of information in different formats, in some areas or domains such as healthcare (e.g. medical records), justice (e.g. sworn declarations), assurance (e.g. declara-tion) and insurance (e.g. Research articles and reports). A lot of this information is stored as narrative texts, hindering its use for the decision making processes. The process of discovering the knowledge contained in these texts, or creating new hypotheses according to them include a lot of time and effort [14], [24] that cannot be afforded by most organizations.Generally, this problem remains a constant challenge due to the difficulty of organizations to absorb and use the information they need. It should be noted that, the limits on the reading speed of a human being make it impossible to capture the key information in a short time when there is a larg...

show abstract

Towards applying text mining and natural language processing for biomedical ontology acquisition

Cited by 12 publications

References 25 publications

Redundancy in electronic health record corpora: analysis, impact on text mining performance and mitigation strategies

Redundancy in electronic health record corpora: analysis, impact on text mining performance and mitigation strategies

Mineração de textos biomédicos: uma revisão bibliométrica

Great Model: A Model for the Automatic Generation of Semantic Relations Between Text Summaries Based on the Relevant Information to the User

Contact Info

Product

Resources

About