Effect of importance sampling on robust segmentation of audio-cough events in noisy environments

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

63 Downloads (Pure)

Abstract

This paper proposes a new cough detection system based on audio signals acquired from conventional smartphones. The system relies on local Hu moments to characterize cough events and a Λ-NN classifier to distinguish cough events from non-cough ones (speech, laugh, sneeze, etc.) and noisy sounds. To deal with the unbalance between classes, we employ Distinct-Borderline2 Synthetic Minority Oversampling Technique and a bespoke cost matrix. The system additionally features a post-processing module to avoid isolated false negatives and, this way, increases sensitivity. Evaluation has been carried out using a database comprising a variety of cough and non-cough events and different types of background noise. In this study, we specifically focused on noise likely to appear when the user is carrying the smartphone in daily activities. Different Signal to Noise Ratio values were tested ranging between -15 and 0 dB. Our experiments confirm that local Hu moments are suitable not only for characterizing cough events but also for coping with noisy environments. Results show a sensitivity of 94.17% and a specificity of 92.16% at -15 dB. Thus, our system shows potential as a reliable and place-ubiquitous monitoring device that helps patients self-manage their own respiratory diseases and avoids unreported or fabricated symptoms.
Original languageEnglish
Title of host publicationIEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016
PublisherIEEE
Pages3740-3744
ISBN (Electronic)978-1-4577-0220-4
ISBN (Print)978-1-4577-0219-8
DOIs
Publication statusPublished - 18 Oct 2016

Publication series

NameAnnual International Conference of the IEEE Engineering in Medicine and Biology Society
PublisherIEEE
ISSN (Print)1557-170X
ISSN (Electronic)1558-4615

Fingerprint

Importance sampling
Smartphones
Pulmonary diseases
Signal to noise ratio
Classifiers
Acoustic waves
Monitoring
Processing
Costs
Experiments

Cite this

Monge-Álvarez, J., Hoyos Barceló, C., Lesso, P., Escudero, J., Dahal, K., & Casaseca, J. P. (2016). Effect of importance sampling on robust segmentation of audio-cough events in noisy environments. In IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016 (pp. 3740-3744). (Annual International Conference of the IEEE Engineering in Medicine and Biology Society). IEEE. https://doi.org/10.1109/EMBC.2016.7591541
Monge-Álvarez, Jesús ; Hoyos Barceló, Carlos ; Lesso, Paul ; Escudero, Javier ; Dahal, Keshav ; Casaseca, Juan Pablo. / Effect of importance sampling on robust segmentation of audio-cough events in noisy environments. IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016 . IEEE, 2016. pp. 3740-3744 (Annual International Conference of the IEEE Engineering in Medicine and Biology Society).
@inproceedings{9375a3e6748b408e93e52901cacebae6,
title = "Effect of importance sampling on robust segmentation of audio-cough events in noisy environments",
abstract = "This paper proposes a new cough detection system based on audio signals acquired from conventional smartphones. The system relies on local Hu moments to characterize cough events and a Λ-NN classifier to distinguish cough events from non-cough ones (speech, laugh, sneeze, etc.) and noisy sounds. To deal with the unbalance between classes, we employ Distinct-Borderline2 Synthetic Minority Oversampling Technique and a bespoke cost matrix. The system additionally features a post-processing module to avoid isolated false negatives and, this way, increases sensitivity. Evaluation has been carried out using a database comprising a variety of cough and non-cough events and different types of background noise. In this study, we specifically focused on noise likely to appear when the user is carrying the smartphone in daily activities. Different Signal to Noise Ratio values were tested ranging between -15 and 0 dB. Our experiments confirm that local Hu moments are suitable not only for characterizing cough events but also for coping with noisy environments. Results show a sensitivity of 94.17{\%} and a specificity of 92.16{\%} at -15 dB. Thus, our system shows potential as a reliable and place-ubiquitous monitoring device that helps patients self-manage their own respiratory diseases and avoids unreported or fabricated symptoms.",
author = "Jes{\'u}s Monge-{\'A}lvarez and {Hoyos Barcel{\'o}}, Carlos and Paul Lesso and Javier Escudero and Keshav Dahal and Casaseca, {Juan Pablo}",
year = "2016",
month = "10",
day = "18",
doi = "10.1109/EMBC.2016.7591541",
language = "English",
isbn = "978-1-4577-0219-8",
series = "Annual International Conference of the IEEE Engineering in Medicine and Biology Society",
publisher = "IEEE",
pages = "3740--3744",
booktitle = "IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016",
address = "United States",

}

Monge-Álvarez, J, Hoyos Barceló, C, Lesso, P, Escudero, J, Dahal, K & Casaseca, JP 2016, Effect of importance sampling on robust segmentation of audio-cough events in noisy environments. in IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016 . Annual International Conference of the IEEE Engineering in Medicine and Biology Society, IEEE, pp. 3740-3744. https://doi.org/10.1109/EMBC.2016.7591541

Effect of importance sampling on robust segmentation of audio-cough events in noisy environments. / Monge-Álvarez, Jesús; Hoyos Barceló, Carlos; Lesso, Paul; Escudero, Javier; Dahal, Keshav; Casaseca, Juan Pablo.

IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016 . IEEE, 2016. p. 3740-3744 (Annual International Conference of the IEEE Engineering in Medicine and Biology Society).

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

TY - GEN

T1 - Effect of importance sampling on robust segmentation of audio-cough events in noisy environments

AU - Monge-Álvarez, Jesús

AU - Hoyos Barceló, Carlos

AU - Lesso, Paul

AU - Escudero, Javier

AU - Dahal, Keshav

AU - Casaseca, Juan Pablo

PY - 2016/10/18

Y1 - 2016/10/18

N2 - This paper proposes a new cough detection system based on audio signals acquired from conventional smartphones. The system relies on local Hu moments to characterize cough events and a Λ-NN classifier to distinguish cough events from non-cough ones (speech, laugh, sneeze, etc.) and noisy sounds. To deal with the unbalance between classes, we employ Distinct-Borderline2 Synthetic Minority Oversampling Technique and a bespoke cost matrix. The system additionally features a post-processing module to avoid isolated false negatives and, this way, increases sensitivity. Evaluation has been carried out using a database comprising a variety of cough and non-cough events and different types of background noise. In this study, we specifically focused on noise likely to appear when the user is carrying the smartphone in daily activities. Different Signal to Noise Ratio values were tested ranging between -15 and 0 dB. Our experiments confirm that local Hu moments are suitable not only for characterizing cough events but also for coping with noisy environments. Results show a sensitivity of 94.17% and a specificity of 92.16% at -15 dB. Thus, our system shows potential as a reliable and place-ubiquitous monitoring device that helps patients self-manage their own respiratory diseases and avoids unreported or fabricated symptoms.

AB - This paper proposes a new cough detection system based on audio signals acquired from conventional smartphones. The system relies on local Hu moments to characterize cough events and a Λ-NN classifier to distinguish cough events from non-cough ones (speech, laugh, sneeze, etc.) and noisy sounds. To deal with the unbalance between classes, we employ Distinct-Borderline2 Synthetic Minority Oversampling Technique and a bespoke cost matrix. The system additionally features a post-processing module to avoid isolated false negatives and, this way, increases sensitivity. Evaluation has been carried out using a database comprising a variety of cough and non-cough events and different types of background noise. In this study, we specifically focused on noise likely to appear when the user is carrying the smartphone in daily activities. Different Signal to Noise Ratio values were tested ranging between -15 and 0 dB. Our experiments confirm that local Hu moments are suitable not only for characterizing cough events but also for coping with noisy environments. Results show a sensitivity of 94.17% and a specificity of 92.16% at -15 dB. Thus, our system shows potential as a reliable and place-ubiquitous monitoring device that helps patients self-manage their own respiratory diseases and avoids unreported or fabricated symptoms.

U2 - 10.1109/EMBC.2016.7591541

DO - 10.1109/EMBC.2016.7591541

M3 - Conference contribution

SN - 978-1-4577-0219-8

T3 - Annual International Conference of the IEEE Engineering in Medicine and Biology Society

SP - 3740

EP - 3744

BT - IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016

PB - IEEE

ER -

Monge-Álvarez J, Hoyos Barceló C, Lesso P, Escudero J, Dahal K, Casaseca JP. Effect of importance sampling on robust segmentation of audio-cough events in noisy environments. In IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016 . IEEE. 2016. p. 3740-3744. (Annual International Conference of the IEEE Engineering in Medicine and Biology Society). https://doi.org/10.1109/EMBC.2016.7591541