A novel generative adversarial networks modelling for the class imbalance problem in high dimensional omics data

Samuel Cusworth; Georgios V. Gkoutos; Animesh Acharjee

doi:10.1186/s12911-024-02487-2

A novel generative adversarial networks modelling for the class imbalance problem in high dimensional omics data

Samuel Cusworth, Georgios V. Gkoutos, Animesh Acharjee^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

67 Downloads (Pure)

Abstract

Class imbalance remains a large problem in high-throughput omics analyses, causing bias towards the over-represented class when training machine learning-based classifiers. Oversampling is a common method used to balance classes, allowing for better generalization of the training data. More naive approaches can introduce other biases into the data, being especially sensitive to inaccuracies in the training data, a problem considering the characteristically noisy data obtained in healthcare. This is especially a problem with high-dimensional data. A generative adversarial network-based method is proposed for creating synthetic samples from small, high-dimensional data, to improve upon other more naive generative approaches. The method was compared with ‘synthetic minority over-sampling technique’ (SMOTE) and ‘random oversampling’ (RO). Generative methods were validated by training classifiers on the balanced data.

Original language	English
Article number	90
Number of pages	17
Journal	BMC Medical Informatics and Decision Making
Volume	24
Issue number	1
DOIs	https://doi.org/10.1186/s12911-024-02487-2
Publication status	Published - 28 Mar 2024

Bibliographical note

Funding:
The authors acknowledge support from the HYPERMARKER (Grant agreement ID 101095480), NIHR Birmingham SRMRC and the MRC Health Data Research UK (HDRUK/CFC/01), an initiative funded by UK Research and Innovation, Department of Health and Social Care (England) and the devolved administrations, and leading medical research charities. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research, the Medical Research Council or the Department of Health.

Keywords

GAN
Multiomics
Synthetic data
Class imbalance

Access to Document

10.1186/s12911-024-02487-2Licence: Creative Commons: Attribution (CC BY)

CusworthS2024novelFinal published version, 1.92 MBLicence: Creative Commons: Attribution (CC BY)

Hypermarker: Personalised pharmacometabolomic optimisation of treatment for hypertension
Acharjee, A. (Co-Investigator), Gkoutos, G. (Co-Investigator), Mobley, A. (Co-Investigator) & Kotecha, D. (Principal Investigator)
UKRI Horizon Europe Underwriting â€“ Innovate UK
1/01/23 → 31/12/26
Project: Research

Cite this

@article{1f6075b845c94b71b37018cfcf3b08cd,

title = "A novel generative adversarial networks modelling for the class imbalance problem in high dimensional omics data",

abstract = "Class imbalance remains a large problem in high-throughput omics analyses, causing bias towards the over-represented class when training machine learning-based classifiers. Oversampling is a common method used to balance classes, allowing for better generalization of the training data. More naive approaches can introduce other biases into the data, being especially sensitive to inaccuracies in the training data, a problem considering the characteristically noisy data obtained in healthcare. This is especially a problem with high-dimensional data. A generative adversarial network-based method is proposed for creating synthetic samples from small, high-dimensional data, to improve upon other more naive generative approaches. The method was compared with {\textquoteleft}synthetic minority over-sampling technique{\textquoteright} (SMOTE) and {\textquoteleft}random oversampling{\textquoteright} (RO). Generative methods were validated by training classifiers on the balanced data.",

keywords = "GAN, Multiomics, Synthetic data, Class imbalance",

author = "Samuel Cusworth and Gkoutos, {Georgios V.} and Animesh Acharjee",

note = "Funding: The authors acknowledge support from the HYPERMARKER (Grant agreement ID 101095480), NIHR Birmingham SRMRC and the MRC Health Data Research UK (HDRUK/CFC/01), an initiative funded by UK Research and Innovation, Department of Health and Social Care (England) and the devolved administrations, and leading medical research charities. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research, the Medical Research Council or the Department of Health.",

year = "2024",

month = mar,

day = "28",

doi = "10.1186/s12911-024-02487-2",

language = "English",

volume = "24",

journal = "BMC Medical Informatics and Decision Making",

issn = "1472-6947",

publisher = "Springer",

number = "1",

}

TY - JOUR

T1 - A novel generative adversarial networks modelling for the class imbalance problem in high dimensional omics data

AU - Cusworth, Samuel

AU - Gkoutos, Georgios V.

AU - Acharjee, Animesh

N1 - Funding: The authors acknowledge support from the HYPERMARKER (Grant agreement ID 101095480), NIHR Birmingham SRMRC and the MRC Health Data Research UK (HDRUK/CFC/01), an initiative funded by UK Research and Innovation, Department of Health and Social Care (England) and the devolved administrations, and leading medical research charities. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research, the Medical Research Council or the Department of Health.

PY - 2024/3/28

Y1 - 2024/3/28

N2 - Class imbalance remains a large problem in high-throughput omics analyses, causing bias towards the over-represented class when training machine learning-based classifiers. Oversampling is a common method used to balance classes, allowing for better generalization of the training data. More naive approaches can introduce other biases into the data, being especially sensitive to inaccuracies in the training data, a problem considering the characteristically noisy data obtained in healthcare. This is especially a problem with high-dimensional data. A generative adversarial network-based method is proposed for creating synthetic samples from small, high-dimensional data, to improve upon other more naive generative approaches. The method was compared with ‘synthetic minority over-sampling technique’ (SMOTE) and ‘random oversampling’ (RO). Generative methods were validated by training classifiers on the balanced data.

AB - Class imbalance remains a large problem in high-throughput omics analyses, causing bias towards the over-represented class when training machine learning-based classifiers. Oversampling is a common method used to balance classes, allowing for better generalization of the training data. More naive approaches can introduce other biases into the data, being especially sensitive to inaccuracies in the training data, a problem considering the characteristically noisy data obtained in healthcare. This is especially a problem with high-dimensional data. A generative adversarial network-based method is proposed for creating synthetic samples from small, high-dimensional data, to improve upon other more naive generative approaches. The method was compared with ‘synthetic minority over-sampling technique’ (SMOTE) and ‘random oversampling’ (RO). Generative methods were validated by training classifiers on the balanced data.

KW - GAN

KW - Multiomics

KW - Synthetic data

KW - Class imbalance

U2 - 10.1186/s12911-024-02487-2

DO - 10.1186/s12911-024-02487-2

M3 - Article

C2 - 38549123

SN - 1472-6947

VL - 24

JO - BMC Medical Informatics and Decision Making

JF - BMC Medical Informatics and Decision Making

IS - 1

M1 - 90

ER -

A novel generative adversarial networks modelling for the class imbalance problem in high dimensional omics data

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Projects

Hypermarker: Personalised pharmacometabolomic optimisation of treatment for hypertension

Cite this