Adverse drug reaction classification with deep neural networks

Trung Huynh, Yulan He, Allistair Willis, Stefan Rueger

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

Abstract

We study the problem of detecting sentences describing adverse drug reactions (ADRs) and frame the problem as binary classification. We investigate different neural network (NN) architectures for ADR classification. In particular, we propose two new neural network models, Convolutional Recurrent Neural Network (CRNN) by concatenating convolutional neural networks with recurrent neural networks, and Convolutional Neural Network with Attention (CNNA) by adding attention weights into convolutional neural networks. We evaluate various NN architectures on a Twitter dataset containing informal language and an Adverse Drug Effects (ADE) dataset constructed by sampling from MEDLINE case reports. Experimental results show that all the NN architectures outperform the traditional maximum entropy classifiers trained from n-grams with different weighting strategies considerably on both datasets. On the Twitter dataset, all the NN architectures perform similarly. But on the ADE dataset, CNN performs better than other more complex CNN variants. Nevertheless, CNNA allows the visualisation of attention weights of words when making classification decisions and hence is more appropriate for the extraction of word subsequences describing ADRs.
Original languageEnglish
Title of host publicationProceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers
Pages877–887
Number of pages11
Publication statusPublished - 16 Dec 2016
Event26th International Conference on Computational Linguistics: COLIN 2016 - Osaka, Japan
Duration: 11 Dec 201616 Dec 2016

Conference

Conference26th International Conference on Computational Linguistics
CountryJapan
CityOsaka
Period11/12/1616/12/16

Fingerprint

Neural networks
Network architecture
Recurrent neural networks
Deep neural networks
Classifiers
Entropy
Visualization
Sampling

Bibliographical note

This work is licensed under a Creative Commons Attribution 4.0 International Licence http://creativecommons.org/licenses/by/4.0/

Funding: EPSRC AMR4AMR project EP/M02735X/1).

Cite this

Huynh, T., He, Y., Willis, A., & Rueger, S. (2016). Adverse drug reaction classification with deep neural networks. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers (pp. 877–887)
Huynh, Trung ; He, Yulan ; Willis, Allistair ; Rueger, Stefan. / Adverse drug reaction classification with deep neural networks. Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers. 2016. pp. 877–887
@inproceedings{7701a50a61874eeaa605100ccc1fd25b,
title = "Adverse drug reaction classification with deep neural networks",
abstract = "We study the problem of detecting sentences describing adverse drug reactions (ADRs) and frame the problem as binary classification. We investigate different neural network (NN) architectures for ADR classification. In particular, we propose two new neural network models, Convolutional Recurrent Neural Network (CRNN) by concatenating convolutional neural networks with recurrent neural networks, and Convolutional Neural Network with Attention (CNNA) by adding attention weights into convolutional neural networks. We evaluate various NN architectures on a Twitter dataset containing informal language and an Adverse Drug Effects (ADE) dataset constructed by sampling from MEDLINE case reports. Experimental results show that all the NN architectures outperform the traditional maximum entropy classifiers trained from n-grams with different weighting strategies considerably on both datasets. On the Twitter dataset, all the NN architectures perform similarly. But on the ADE dataset, CNN performs better than other more complex CNN variants. Nevertheless, CNNA allows the visualisation of attention weights of words when making classification decisions and hence is more appropriate for the extraction of word subsequences describing ADRs.",
author = "Trung Huynh and Yulan He and Allistair Willis and Stefan Rueger",
note = "This work is licensed under a Creative Commons Attribution 4.0 International Licence http://creativecommons.org/licenses/by/4.0/ Funding: EPSRC AMR4AMR project EP/M02735X/1).",
year = "2016",
month = "12",
day = "16",
language = "English",
pages = "877–887",
booktitle = "Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers",

}

Huynh, T, He, Y, Willis, A & Rueger, S 2016, Adverse drug reaction classification with deep neural networks. in Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers. pp. 877–887, 26th International Conference on Computational Linguistics, Osaka, Japan, 11/12/16.

Adverse drug reaction classification with deep neural networks. / Huynh, Trung; He, Yulan; Willis, Allistair; Rueger, Stefan.

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers. 2016. p. 877–887.

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

TY - GEN

T1 - Adverse drug reaction classification with deep neural networks

AU - Huynh, Trung

AU - He, Yulan

AU - Willis, Allistair

AU - Rueger, Stefan

N1 - This work is licensed under a Creative Commons Attribution 4.0 International Licence http://creativecommons.org/licenses/by/4.0/ Funding: EPSRC AMR4AMR project EP/M02735X/1).

PY - 2016/12/16

Y1 - 2016/12/16

N2 - We study the problem of detecting sentences describing adverse drug reactions (ADRs) and frame the problem as binary classification. We investigate different neural network (NN) architectures for ADR classification. In particular, we propose two new neural network models, Convolutional Recurrent Neural Network (CRNN) by concatenating convolutional neural networks with recurrent neural networks, and Convolutional Neural Network with Attention (CNNA) by adding attention weights into convolutional neural networks. We evaluate various NN architectures on a Twitter dataset containing informal language and an Adverse Drug Effects (ADE) dataset constructed by sampling from MEDLINE case reports. Experimental results show that all the NN architectures outperform the traditional maximum entropy classifiers trained from n-grams with different weighting strategies considerably on both datasets. On the Twitter dataset, all the NN architectures perform similarly. But on the ADE dataset, CNN performs better than other more complex CNN variants. Nevertheless, CNNA allows the visualisation of attention weights of words when making classification decisions and hence is more appropriate for the extraction of word subsequences describing ADRs.

AB - We study the problem of detecting sentences describing adverse drug reactions (ADRs) and frame the problem as binary classification. We investigate different neural network (NN) architectures for ADR classification. In particular, we propose two new neural network models, Convolutional Recurrent Neural Network (CRNN) by concatenating convolutional neural networks with recurrent neural networks, and Convolutional Neural Network with Attention (CNNA) by adding attention weights into convolutional neural networks. We evaluate various NN architectures on a Twitter dataset containing informal language and an Adverse Drug Effects (ADE) dataset constructed by sampling from MEDLINE case reports. Experimental results show that all the NN architectures outperform the traditional maximum entropy classifiers trained from n-grams with different weighting strategies considerably on both datasets. On the Twitter dataset, all the NN architectures perform similarly. But on the ADE dataset, CNN performs better than other more complex CNN variants. Nevertheless, CNNA allows the visualisation of attention weights of words when making classification decisions and hence is more appropriate for the extraction of word subsequences describing ADRs.

M3 - Conference contribution

SP - 877

EP - 887

BT - Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers

ER -

Huynh T, He Y, Willis A, Rueger S. Adverse drug reaction classification with deep neural networks. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers. 2016. p. 877–887