Classification of oximetry signals using Bayesian neural networks to assist in the detection of obstructive sleep apnoea syndrome

J.V. Marcos, R. Homero, D. Álvarez, Ian T. Nabney, F. del Campo, C. Zamarron

Research output: Contribution to journalArticle

Abstract

In the present study, multilayer perceptron (MLP) neural networks were applied to help in the diagnosis of obstructive sleep apnoea syndrome (OSAS). Oxygen saturation (SaO2) recordings from nocturnal pulse oximetry were used for this purpose. We performed time and spectral analysis of these signals to extract 14 features related to OSAS. The performance of two different MLP classifiers was compared: maximum likelihood (ML) and Bayesian (BY) MLP networks. A total of 187 subjects suspected of suffering from OSAS took part in the study. Their SaO2 signals were divided into a training set with 74 recordings and a test set with 113 recordings. BY-MLP networks achieved the best performance on the test set with 85.58% accuracy (87.76% sensitivity and 82.39% specificity). These results were substantially better than those provided by ML-MLP networks, which were affected by overfitting and achieved an accuracy of 76.81% (86.42% sensitivity and 62.83% specificity). Our results suggest that the Bayesian framework is preferred to implement our MLP classifiers. The proposed BY-MLP networks could be used for early OSAS detection. They could contribute to overcome the difficulties of nocturnal polysomnography (PSG) and thus reduce the demand for these studies.
Original languageEnglish
Pages (from-to)375-394
Number of pages20
JournalPhysiological Measurement
Volume31
Issue number3
DOIs
Publication statusPublished - 2010

Fingerprint

Neural Networks (Computer)
Oximetry
Obstructive Sleep Apnea
Multilayer neural networks
Neural networks
Maximum likelihood
Classifiers
Sensitivity and Specificity
Polysomnography
Sleep
Spectrum analysis
Oxygen

Bibliographical note

Copyright of the Institute of Physics

Keywords

  • obstructive sleep apnoea syndrome (OSAS)
  • nocturnal pulse oximetry
  • multilayer perceptron (MLP)
  • maximum likelihood
  • Bayesian inference

Cite this

Marcos, J.V. ; Homero, R. ; Álvarez, D. ; Nabney, Ian T. ; del Campo, F. ; Zamarron, C. / Classification of oximetry signals using Bayesian neural networks to assist in the detection of obstructive sleep apnoea syndrome. In: Physiological Measurement. 2010 ; Vol. 31, No. 3. pp. 375-394.
@article{9fec0582d95a48338ad1b1c5b0427470,
title = "Classification of oximetry signals using Bayesian neural networks to assist in the detection of obstructive sleep apnoea syndrome",
abstract = "In the present study, multilayer perceptron (MLP) neural networks were applied to help in the diagnosis of obstructive sleep apnoea syndrome (OSAS). Oxygen saturation (SaO2) recordings from nocturnal pulse oximetry were used for this purpose. We performed time and spectral analysis of these signals to extract 14 features related to OSAS. The performance of two different MLP classifiers was compared: maximum likelihood (ML) and Bayesian (BY) MLP networks. A total of 187 subjects suspected of suffering from OSAS took part in the study. Their SaO2 signals were divided into a training set with 74 recordings and a test set with 113 recordings. BY-MLP networks achieved the best performance on the test set with 85.58{\%} accuracy (87.76{\%} sensitivity and 82.39{\%} specificity). These results were substantially better than those provided by ML-MLP networks, which were affected by overfitting and achieved an accuracy of 76.81{\%} (86.42{\%} sensitivity and 62.83{\%} specificity). Our results suggest that the Bayesian framework is preferred to implement our MLP classifiers. The proposed BY-MLP networks could be used for early OSAS detection. They could contribute to overcome the difficulties of nocturnal polysomnography (PSG) and thus reduce the demand for these studies.",
keywords = "obstructive sleep apnoea syndrome (OSAS), nocturnal pulse oximetry, multilayer perceptron (MLP), maximum likelihood, Bayesian inference",
author = "J.V. Marcos and R. Homero and D. {\'A}lvarez and Nabney, {Ian T.} and {del Campo}, F. and C. Zamarron",
note = "Copyright of the Institute of Physics",
year = "2010",
doi = "10.1088/0967-3334/31/3/007",
language = "English",
volume = "31",
pages = "375--394",
journal = "Physiological Measurement",
issn = "0967-3334",
publisher = "IOP Publishing Ltd.",
number = "3",

}

Classification of oximetry signals using Bayesian neural networks to assist in the detection of obstructive sleep apnoea syndrome. / Marcos, J.V.; Homero, R.; Álvarez, D.; Nabney, Ian T.; del Campo, F.; Zamarron, C.

In: Physiological Measurement, Vol. 31, No. 3, 2010, p. 375-394.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Classification of oximetry signals using Bayesian neural networks to assist in the detection of obstructive sleep apnoea syndrome

AU - Marcos, J.V.

AU - Homero, R.

AU - Álvarez, D.

AU - Nabney, Ian T.

AU - del Campo, F.

AU - Zamarron, C.

N1 - Copyright of the Institute of Physics

PY - 2010

Y1 - 2010

N2 - In the present study, multilayer perceptron (MLP) neural networks were applied to help in the diagnosis of obstructive sleep apnoea syndrome (OSAS). Oxygen saturation (SaO2) recordings from nocturnal pulse oximetry were used for this purpose. We performed time and spectral analysis of these signals to extract 14 features related to OSAS. The performance of two different MLP classifiers was compared: maximum likelihood (ML) and Bayesian (BY) MLP networks. A total of 187 subjects suspected of suffering from OSAS took part in the study. Their SaO2 signals were divided into a training set with 74 recordings and a test set with 113 recordings. BY-MLP networks achieved the best performance on the test set with 85.58% accuracy (87.76% sensitivity and 82.39% specificity). These results were substantially better than those provided by ML-MLP networks, which were affected by overfitting and achieved an accuracy of 76.81% (86.42% sensitivity and 62.83% specificity). Our results suggest that the Bayesian framework is preferred to implement our MLP classifiers. The proposed BY-MLP networks could be used for early OSAS detection. They could contribute to overcome the difficulties of nocturnal polysomnography (PSG) and thus reduce the demand for these studies.

AB - In the present study, multilayer perceptron (MLP) neural networks were applied to help in the diagnosis of obstructive sleep apnoea syndrome (OSAS). Oxygen saturation (SaO2) recordings from nocturnal pulse oximetry were used for this purpose. We performed time and spectral analysis of these signals to extract 14 features related to OSAS. The performance of two different MLP classifiers was compared: maximum likelihood (ML) and Bayesian (BY) MLP networks. A total of 187 subjects suspected of suffering from OSAS took part in the study. Their SaO2 signals were divided into a training set with 74 recordings and a test set with 113 recordings. BY-MLP networks achieved the best performance on the test set with 85.58% accuracy (87.76% sensitivity and 82.39% specificity). These results were substantially better than those provided by ML-MLP networks, which were affected by overfitting and achieved an accuracy of 76.81% (86.42% sensitivity and 62.83% specificity). Our results suggest that the Bayesian framework is preferred to implement our MLP classifiers. The proposed BY-MLP networks could be used for early OSAS detection. They could contribute to overcome the difficulties of nocturnal polysomnography (PSG) and thus reduce the demand for these studies.

KW - obstructive sleep apnoea syndrome (OSAS)

KW - nocturnal pulse oximetry

KW - multilayer perceptron (MLP)

KW - maximum likelihood

KW - Bayesian inference

UR - http://www.scopus.com/inward/record.url?scp=77149120423&partnerID=8YFLogxK

U2 - 10.1088/0967-3334/31/3/007

DO - 10.1088/0967-3334/31/3/007

M3 - Article

VL - 31

SP - 375

EP - 394

JO - Physiological Measurement

JF - Physiological Measurement

SN - 0967-3334

IS - 3

ER -