Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions

A. Kolchinsky, A. Lourenço, L. Li, L. M. Rocha

Research output: Chapter in Book/Report/Conference proceedingConference contribution

10 Citations (Scopus)

Abstract

Background. Drug-drug interaction (DDI) is a major cause of morbidity and mortality. DDI research includes the study of different aspects of drug interactions, from in vitro pharmacology, which deals with drug interaction mechanisms, to pharmaco-epidemiology, which investigates the effects of DDI on drug efficacy and adverse drug reactions. Biomedical literature mining can aid both kinds of approaches by extracting relevant DDI signals from either the published literature or large clinical databases. However, though drug interaction is an ideal area for translational research, the inclusion of literature mining methodologies in DDI workflows is still very preliminary. One area that can benefit from literature mining is the automatic identification of a large number of potential DDIs, whose pharmacological mechanisms and clinical significance can then be studied via in vitro pharmacology and in populo pharmaco-epidemiology. Experiments. We implemented a set of classifiers for identifying published articles relevant to experimental pharmacokinetic DDI evidence. These documents are important for identifying causal mechanisms behind putative drug-drug interactions, an important step in the extraction of large numbers of potential DDIs. We evaluate performance of several linear classifiers on PubMed abstracts, under different feature transformation and dimensionality reduction methods. In addition, we investigate the performance benefits of including various publicly-available named entity recognition features, as well as a set of internally-developed pharmacokinetic dictionaries. Results. We found that several classifiers performed well in distinguishing relevant and irrelevant abstracts. We found that the combination of unigram and bigram textual features gave better performance than unigram features alone, and also that normalization transforms that adjusted for feature frequency and document length improved classification. For some classifiers, such as linear discriminant analysis (LDA), proper dimensionality reduction had a large impact on performance. Finally, the inclusion of NER features and dictionaries was found not to help classification.

Original languageEnglish (US)
Title of host publication18th Pacific Symposium on Biocomputing, PSB 2013
PublisherWorld Scientific Publishing Co. Pte Ltd
Pages409-420
Number of pages12
ISBN (Electronic)9781627480161
StatePublished - 2013
Event18th Pacific Symposium on Biocomputing, PSB 2013 - Kohala Coast, United States
Duration: Jan 3 2013Jan 7 2013

Other

Other18th Pacific Symposium on Biocomputing, PSB 2013
CountryUnited States
CityKohala Coast
Period1/3/131/7/13

Fingerprint

Drug interactions
Pharmacokinetics
Drug Interactions
Classifiers
Pharmaceutical Preparations
Epidemiology
Pharmacology
Glossaries
Translational Medical Research
Workflow
Discriminant Analysis
Discriminant analysis
Drug-Related Side Effects and Adverse Reactions
PubMed

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Biomedical Engineering
  • Medicine(all)

Cite this

Kolchinsky, A., Lourenço, A., Li, L., & Rocha, L. M. (2013). Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions. In 18th Pacific Symposium on Biocomputing, PSB 2013 (pp. 409-420). World Scientific Publishing Co. Pte Ltd.

Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions. / Kolchinsky, A.; Lourenço, A.; Li, L.; Rocha, L. M.

18th Pacific Symposium on Biocomputing, PSB 2013. World Scientific Publishing Co. Pte Ltd, 2013. p. 409-420.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kolchinsky, A, Lourenço, A, Li, L & Rocha, LM 2013, Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions. in 18th Pacific Symposium on Biocomputing, PSB 2013. World Scientific Publishing Co. Pte Ltd, pp. 409-420, 18th Pacific Symposium on Biocomputing, PSB 2013, Kohala Coast, United States, 1/3/13.
Kolchinsky A, Lourenço A, Li L, Rocha LM. Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions. In 18th Pacific Symposium on Biocomputing, PSB 2013. World Scientific Publishing Co. Pte Ltd. 2013. p. 409-420
Kolchinsky, A. ; Lourenço, A. ; Li, L. ; Rocha, L. M. / Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions. 18th Pacific Symposium on Biocomputing, PSB 2013. World Scientific Publishing Co. Pte Ltd, 2013. pp. 409-420
@inproceedings{1bc3040a900a4cbea0dd6b48b19b70e3,
title = "Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions",
abstract = "Background. Drug-drug interaction (DDI) is a major cause of morbidity and mortality. DDI research includes the study of different aspects of drug interactions, from in vitro pharmacology, which deals with drug interaction mechanisms, to pharmaco-epidemiology, which investigates the effects of DDI on drug efficacy and adverse drug reactions. Biomedical literature mining can aid both kinds of approaches by extracting relevant DDI signals from either the published literature or large clinical databases. However, though drug interaction is an ideal area for translational research, the inclusion of literature mining methodologies in DDI workflows is still very preliminary. One area that can benefit from literature mining is the automatic identification of a large number of potential DDIs, whose pharmacological mechanisms and clinical significance can then be studied via in vitro pharmacology and in populo pharmaco-epidemiology. Experiments. We implemented a set of classifiers for identifying published articles relevant to experimental pharmacokinetic DDI evidence. These documents are important for identifying causal mechanisms behind putative drug-drug interactions, an important step in the extraction of large numbers of potential DDIs. We evaluate performance of several linear classifiers on PubMed abstracts, under different feature transformation and dimensionality reduction methods. In addition, we investigate the performance benefits of including various publicly-available named entity recognition features, as well as a set of internally-developed pharmacokinetic dictionaries. Results. We found that several classifiers performed well in distinguishing relevant and irrelevant abstracts. We found that the combination of unigram and bigram textual features gave better performance than unigram features alone, and also that normalization transforms that adjusted for feature frequency and document length improved classification. For some classifiers, such as linear discriminant analysis (LDA), proper dimensionality reduction had a large impact on performance. Finally, the inclusion of NER features and dictionaries was found not to help classification.",
author = "A. Kolchinsky and A. Louren{\cc}o and L. Li and Rocha, {L. M.}",
year = "2013",
language = "English (US)",
pages = "409--420",
booktitle = "18th Pacific Symposium on Biocomputing, PSB 2013",
publisher = "World Scientific Publishing Co. Pte Ltd",

}

TY - GEN

T1 - Evaluation of linear classifiers on articles containing pharmacokinetic evidence of drug-drug interactions

AU - Kolchinsky, A.

AU - Lourenço, A.

AU - Li, L.

AU - Rocha, L. M.

PY - 2013

Y1 - 2013

N2 - Background. Drug-drug interaction (DDI) is a major cause of morbidity and mortality. DDI research includes the study of different aspects of drug interactions, from in vitro pharmacology, which deals with drug interaction mechanisms, to pharmaco-epidemiology, which investigates the effects of DDI on drug efficacy and adverse drug reactions. Biomedical literature mining can aid both kinds of approaches by extracting relevant DDI signals from either the published literature or large clinical databases. However, though drug interaction is an ideal area for translational research, the inclusion of literature mining methodologies in DDI workflows is still very preliminary. One area that can benefit from literature mining is the automatic identification of a large number of potential DDIs, whose pharmacological mechanisms and clinical significance can then be studied via in vitro pharmacology and in populo pharmaco-epidemiology. Experiments. We implemented a set of classifiers for identifying published articles relevant to experimental pharmacokinetic DDI evidence. These documents are important for identifying causal mechanisms behind putative drug-drug interactions, an important step in the extraction of large numbers of potential DDIs. We evaluate performance of several linear classifiers on PubMed abstracts, under different feature transformation and dimensionality reduction methods. In addition, we investigate the performance benefits of including various publicly-available named entity recognition features, as well as a set of internally-developed pharmacokinetic dictionaries. Results. We found that several classifiers performed well in distinguishing relevant and irrelevant abstracts. We found that the combination of unigram and bigram textual features gave better performance than unigram features alone, and also that normalization transforms that adjusted for feature frequency and document length improved classification. For some classifiers, such as linear discriminant analysis (LDA), proper dimensionality reduction had a large impact on performance. Finally, the inclusion of NER features and dictionaries was found not to help classification.

AB - Background. Drug-drug interaction (DDI) is a major cause of morbidity and mortality. DDI research includes the study of different aspects of drug interactions, from in vitro pharmacology, which deals with drug interaction mechanisms, to pharmaco-epidemiology, which investigates the effects of DDI on drug efficacy and adverse drug reactions. Biomedical literature mining can aid both kinds of approaches by extracting relevant DDI signals from either the published literature or large clinical databases. However, though drug interaction is an ideal area for translational research, the inclusion of literature mining methodologies in DDI workflows is still very preliminary. One area that can benefit from literature mining is the automatic identification of a large number of potential DDIs, whose pharmacological mechanisms and clinical significance can then be studied via in vitro pharmacology and in populo pharmaco-epidemiology. Experiments. We implemented a set of classifiers for identifying published articles relevant to experimental pharmacokinetic DDI evidence. These documents are important for identifying causal mechanisms behind putative drug-drug interactions, an important step in the extraction of large numbers of potential DDIs. We evaluate performance of several linear classifiers on PubMed abstracts, under different feature transformation and dimensionality reduction methods. In addition, we investigate the performance benefits of including various publicly-available named entity recognition features, as well as a set of internally-developed pharmacokinetic dictionaries. Results. We found that several classifiers performed well in distinguishing relevant and irrelevant abstracts. We found that the combination of unigram and bigram textual features gave better performance than unigram features alone, and also that normalization transforms that adjusted for feature frequency and document length improved classification. For some classifiers, such as linear discriminant analysis (LDA), proper dimensionality reduction had a large impact on performance. Finally, the inclusion of NER features and dictionaries was found not to help classification.

UR - http://www.scopus.com/inward/record.url?scp=84887608053&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84887608053&partnerID=8YFLogxK

M3 - Conference contribution

SP - 409

EP - 420

BT - 18th Pacific Symposium on Biocomputing, PSB 2013

PB - World Scientific Publishing Co. Pte Ltd

ER -