Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19

Christian Salamea-Palacios; Tarquino Sánchez-Almeida; Xavier Calderón-Hinojosa; Javier Guaña-Moya; Paulo Castañeda-Romero; Jessica Reina-Trávez

doi:10.18687/LACCEI2022.1.1.145

Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19

Translated title of the contribution: Machine learning models to characterize the cough signal of patients with COVID-19

Christian Salamea-Palacios, Tarquino Sánchez-Almeida, Xavier Calderón-Hinojosa, Javier Guaña-Moya, Paulo Castañeda-Romero, Jessica Reina-Trávez

Research Group on Interaction, Robotics and Automatics (GIIRA)

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Automatic recognition of audio signals is a challenging signal task due to the difficulty of extracting important attributes from such signals, which relies heavily on discriminating acoustic features to determine the type of cough audio coming from COVID-19 patients. In this work, the use of state-of-the-art pre-trained models and a convolutional neural network for the extraction of characteristics of a cough signal from patients with COVID-19 is analyzed. A comparison of three machine learning models has been proposed to extract the features containing relevant information, leading to the recognition of the COVID-19 cough signal. The first model is based on a basic convolutional neural network, the second is based on a YAMNet pre-treatment model, and the third is a VGGish pre-trained model. The experimental results carried out with a ComPare 2021 CCS database show that models, of the three, used, VGGish to provide better performance when extracting the characteristics of the audio signals of the COVID-19 cough signal, having as results the performance metrics f1 score and accuracy with values of 30.76% and 80.51%, representing an improvement of 6.06% and 3.61% compared to the YANMet model, and the confusion matrices, which validate the mentioned model.

Translated title of the contribution	Machine learning models to characterize the cough signal of patients with COVID-19
Original language	Spanish
Title of host publication	20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology
Subtitle of host publication	"Education, Research and Leadership in Post-Pandemic Engineering: Resilient Inclusive and Sustainable Actions", LACCEI 2022
Editors	Maria M. Larrondo Petrie, Jose Texier, Andrea Pena, Jose Angel Sanchez Viloria
Publisher	Latin American and Caribbean Consortium of Engineering Institutions
ISBN (Electronic)	9786289520705
DOIs	https://doi.org/10.18687/LACCEI2022.1.1.145
State	Published - 2022
Event	20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology, LACCEI 2022 - Boca Raton, United States Duration: 18 Jul 2022 → 22 Jul 2022

Publication series

Name	Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology
Volume	2022-July
ISSN (Electronic)	2414-6390

Conference

Conference	20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology, LACCEI 2022
Country/Territory	United States
City	Boca Raton
Period	18/07/22 → 22/07/22

Bibliographical note

Funding Information:
A la Corporación Ecuatoriana para el Desarrollo de la Investigación y Academia, CEDIA, por el financiamiento brindado a la investigación, desarrollo e innovación a través de los proyectos CEPRA, en especial el proyecto CEPRA-XV-2021-011: Caracterización de la tos provocada por el COVID-19 en pacientes con diagnóstico positivo. Los autores agradecen a la Escuela Politécnica Nacional, la Universidad Politécnica Salesiana y la Pontificia Universidad Católica del Ecuador.

Publisher Copyright:
© 2022 Latin American and Caribbean Consortium of Engineering Institutions. All rights reserved.

Access to Document

10.18687/LACCEI2022.1.1.145

Cite this

Salamea-Palacios, C., Sánchez-Almeida, T., Calderón-Hinojosa, X., Guaña-Moya, J., Castañeda-Romero, P., & Reina-Trávez, J. (2022). Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19. In M. M. Larrondo Petrie, J. Texier, A. Pena, & J. A. S. Viloria (Eds.), 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology: "Education, Research and Leadership in Post-Pandemic Engineering: Resilient Inclusive and Sustainable Actions", LACCEI 2022 (Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology; Vol. 2022-July). Latin American and Caribbean Consortium of Engineering Institutions. https://doi.org/10.18687/LACCEI2022.1.1.145

Salamea-Palacios, Christian ; Sánchez-Almeida, Tarquino ; Calderón-Hinojosa, Xavier et al. / Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19. 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology: "Education, Research and Leadership in Post-Pandemic Engineering: Resilient Inclusive and Sustainable Actions", LACCEI 2022. editor / Maria M. Larrondo Petrie ; Jose Texier ; Andrea Pena ; Jose Angel Sanchez Viloria. Latin American and Caribbean Consortium of Engineering Institutions, 2022. (Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology).

@inproceedings{ed73d423753547bcbea023faa31d086e,

title = "Modelos de aprendizaje autom{\'a}tico para caracterizar la se{\~n}al de la tos de pacientes con COVID-19",

abstract = "Automatic recognition of audio signals is a challenging signal task due to the difficulty of extracting important attributes from such signals, which relies heavily on discriminating acoustic features to determine the type of cough audio coming from COVID-19 patients. In this work, the use of state-of-the-art pre-trained models and a convolutional neural network for the extraction of characteristics of a cough signal from patients with COVID-19 is analyzed. A comparison of three machine learning models has been proposed to extract the features containing relevant information, leading to the recognition of the COVID-19 cough signal. The first model is based on a basic convolutional neural network, the second is based on a YAMNet pre-treatment model, and the third is a VGGish pre-trained model. The experimental results carried out with a ComPare 2021 CCS database show that models, of the three, used, VGGish to provide better performance when extracting the characteristics of the audio signals of the COVID-19 cough signal, having as results the performance metrics f1 score and accuracy with values of 30.76% and 80.51%, representing an improvement of 6.06% and 3.61% compared to the YANMet model, and the confusion matrices, which validate the mentioned model.",

keywords = "characterization, Convolutional Neural Network, COVID-19, Transfer Learning, Vggish, YAMNet",

author = "Christian Salamea-Palacios and Tarquino S{\'a}nchez-Almeida and Xavier Calder{\'o}n-Hinojosa and Javier Gua{\~n}a-Moya and Paulo Casta{\~n}eda-Romero and Jessica Reina-Tr{\'a}vez",

note = "Publisher Copyright: {\textcopyright} 2022 Latin American and Caribbean Consortium of Engineering Institutions. All rights reserved.; 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology, LACCEI 2022 ; Conference date: 18-07-2022 Through 22-07-2022",

year = "2022",

doi = "10.18687/LACCEI2022.1.1.145",

language = "Espa{\~n}ol",

series = "Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology",

publisher = "Latin American and Caribbean Consortium of Engineering Institutions",

editor = "{Larrondo Petrie}, {Maria M.} and Jose Texier and Andrea Pena and Viloria, {Jose Angel Sanchez}",

booktitle = "20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology",

}

Salamea-Palacios, C, Sánchez-Almeida, T, Calderón-Hinojosa, X, Guaña-Moya, J, Castañeda-Romero, P & Reina-Trávez, J 2022, Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19. in MM Larrondo Petrie, J Texier, A Pena & JAS Viloria (eds), 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology: "Education, Research and Leadership in Post-Pandemic Engineering: Resilient Inclusive and Sustainable Actions", LACCEI 2022. Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology, vol. 2022-July, Latin American and Caribbean Consortium of Engineering Institutions, 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology, LACCEI 2022, Boca Raton, United States, 18/07/22. https://doi.org/10.18687/LACCEI2022.1.1.145

Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19. / Salamea-Palacios, Christian; Sánchez-Almeida, Tarquino; Calderón-Hinojosa, Xavier et al.
20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology: "Education, Research and Leadership in Post-Pandemic Engineering: Resilient Inclusive and Sustainable Actions", LACCEI 2022. ed. / Maria M. Larrondo Petrie; Jose Texier; Andrea Pena; Jose Angel Sanchez Viloria. Latin American and Caribbean Consortium of Engineering Institutions, 2022. (Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology; Vol. 2022-July).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19

AU - Salamea-Palacios, Christian

AU - Sánchez-Almeida, Tarquino

AU - Calderón-Hinojosa, Xavier

AU - Guaña-Moya, Javier

AU - Castañeda-Romero, Paulo

AU - Reina-Trávez, Jessica

PY - 2022

Y1 - 2022

N2 - Automatic recognition of audio signals is a challenging signal task due to the difficulty of extracting important attributes from such signals, which relies heavily on discriminating acoustic features to determine the type of cough audio coming from COVID-19 patients. In this work, the use of state-of-the-art pre-trained models and a convolutional neural network for the extraction of characteristics of a cough signal from patients with COVID-19 is analyzed. A comparison of three machine learning models has been proposed to extract the features containing relevant information, leading to the recognition of the COVID-19 cough signal. The first model is based on a basic convolutional neural network, the second is based on a YAMNet pre-treatment model, and the third is a VGGish pre-trained model. The experimental results carried out with a ComPare 2021 CCS database show that models, of the three, used, VGGish to provide better performance when extracting the characteristics of the audio signals of the COVID-19 cough signal, having as results the performance metrics f1 score and accuracy with values of 30.76% and 80.51%, representing an improvement of 6.06% and 3.61% compared to the YANMet model, and the confusion matrices, which validate the mentioned model.

AB - Automatic recognition of audio signals is a challenging signal task due to the difficulty of extracting important attributes from such signals, which relies heavily on discriminating acoustic features to determine the type of cough audio coming from COVID-19 patients. In this work, the use of state-of-the-art pre-trained models and a convolutional neural network for the extraction of characteristics of a cough signal from patients with COVID-19 is analyzed. A comparison of three machine learning models has been proposed to extract the features containing relevant information, leading to the recognition of the COVID-19 cough signal. The first model is based on a basic convolutional neural network, the second is based on a YAMNet pre-treatment model, and the third is a VGGish pre-trained model. The experimental results carried out with a ComPare 2021 CCS database show that models, of the three, used, VGGish to provide better performance when extracting the characteristics of the audio signals of the COVID-19 cough signal, having as results the performance metrics f1 score and accuracy with values of 30.76% and 80.51%, representing an improvement of 6.06% and 3.61% compared to the YANMet model, and the confusion matrices, which validate the mentioned model.

KW - characterization

KW - Convolutional Neural Network

KW - COVID-19

KW - Transfer Learning

KW - Vggish

KW - YAMNet

UR - http://www.scopus.com/inward/record.url?scp=85140035951&partnerID=8YFLogxK

U2 - 10.18687/LACCEI2022.1.1.145

DO - 10.18687/LACCEI2022.1.1.145

M3 - Contribución de conferencia

AN - SCOPUS:85140035951

T3 - Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology

BT - 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology

A2 - Larrondo Petrie, Maria M.

A2 - Texier, Jose

A2 - Pena, Andrea

A2 - Viloria, Jose Angel Sanchez

PB - Latin American and Caribbean Consortium of Engineering Institutions

T2 - 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology, LACCEI 2022

Y2 - 18 July 2022 through 22 July 2022

ER -

Salamea-Palacios C, Sánchez-Almeida T, Calderón-Hinojosa X, Guaña-Moya J, Castañeda-Romero P, Reina-Trávez J. Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19. In Larrondo Petrie MM, Texier J, Pena A, Viloria JAS, editors, 20th LACCEI International Multi-Conference for Engineering, Education Caribbean Conference for Engineering and Technology: "Education, Research and Leadership in Post-Pandemic Engineering: Resilient Inclusive and Sustainable Actions", LACCEI 2022. Latin American and Caribbean Consortium of Engineering Institutions. 2022. (Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology). doi: 10.18687/LACCEI2022.1.1.145

Modelos de aprendizaje automático para caracterizar la señal de la tos de pacientes con COVID-19

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this