A Comparative Study of Hybrid Deep Learning Techniques for COVID-19 Detection based on Cough Sound Analysis DOI

Ramya Polaki,

R Annamalai

2022 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS), Journal Year: 2023, Volume and Issue: unknown, P. 478 - 485

Published: Nov. 3, 2023

The COVID-19 pandemic, produced by the SARS-CoV-2 virus, has severe global consequences, resulting in substantial loss of life and posing a serious threat globally. Cough is common sign consequence COVID-19. sound analysis ability to help determine an individual's status. Using deep learning models, this study aims improve accuracy identification based on cough sounds. work employs twelve separate deep-learning models that were extensively trained COUGHVID dataset such as CNN, LSTM, BiLSTM, CNN-LSTM, CNN-BiLSTM with SGD, Adamax optimizer, Attention-based CNN-LSTM Adamax, SGD RMSProp optimizer. To overcome class imbalance, procedures pitch shifting time-frequency masking are used increase positive class. Among these variants, integration attention mechanism model convolutional neural network (CNN), bidirectional long short-term memory (Bi-LSTM) optimizer achieved highest validation accuracy, reaching 95.34%, precision 94.40 %, recall 95.50%, Fl-score 94.44%.

Language: Английский

Towards classification and comprehensive analysis of AI-based COVID-19 diagnostic techniques: A survey DOI

Amna Kosar,

Muhammad Asif, Maaz Bin Ahmad

et al.

Artificial Intelligence in Medicine, Journal Year: 2024, Volume and Issue: 151, P. 102858 - 102858

Published: April 1, 2024

Language: Английский

Citations

4
Reliable data transmission for a VANET-IoIT architecture: A DNN approach DOI
Joydev Ghosh, Neeraj Kumar, Khaled A. Al-Utaibi

et al.

Internet of Things, Journal Year: 2024, Volume and Issue: 25, P. 101129 - 101129

Published: Feb. 18, 2024

Language: Английский

Citations

2
Screening COVID-19 by Swaasa AI platform using cough sounds: a cross-sectional study DOI Creative Commons

Padmalatha Pentakota,

Gowrisree Rudraraju,

Narayana Rao Sripada

et al.

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: Oct. 25, 2023

Abstract The Advent of Artificial Intelligence (AI) has led to the use auditory data for detecting various diseases, including COVID-19. SARS-CoV-2 infection claimed more than six million lives date and therefore, needs a robust screening technique control disease spread. In present study we created validated Swaasa AI platform, which uses signature cough sound symptoms presented by patients screen prioritize COVID-19 patients. We collected from 234 suspects validate our Convolutional Neural Network (CNN) architecture Feedforward (FFANN) (tabular features) based algorithm. final output both models was combined predict likelihood having disease. During clinical validation phase, model showed 75.54% accuracy rate in likely presence COVID-19, with 95.45% sensitivity 73.46% specificity. conducted pilot testing on 183 presumptive COVID subjects, 58 were truly positive, resulting Positive Predictive Value 70.73%. Due high cost technical expertise required currently available rapid methods, there is need cost-effective remote monitoring tool that can serve as preliminary method potential subjects. Therefore, would be highly beneficial could have significant impact reducing its

Language: Английский

Citations

2
A multimodal educational robots driven via dynamic attention DOI Creative Commons

An Jianliang

Frontiers in Neurorobotics, Journal Year: 2024, Volume and Issue: 18

Published: Oct. 31, 2024

With the development of artificial intelligence and robotics technology, application educational robots in teaching is becoming increasingly popular. However, effectively evaluating optimizing multimodal remains a challenge.

Language: Английский

Citations

0
Assessing Data-Driven of Discriminative Deep Learning Models in Classification Task Using Synthetic Pandemic Dataset DOI
Sunday Adeola Ajagbe, Pragasen Mudali, Matthew O. Adigun

et al.

Communications in computer and information science, Journal Year: 2024, Volume and Issue: unknown, P. 282 - 299

Published: Nov. 26, 2024

Language: Английский

Citations

0
Covid-19 Detection from Cough, Breath, And Speech Sounds with Short-Time Fourier Transform and a CNN Model DOI

Ahmet Ekiz,

Kaplan Kaplan

2022 Innovations in Intelligent Systems and Applications Conference (ASYU), Journal Year: 2023, Volume and Issue: unknown, P. 1 - 5

Published: Oct. 11, 2023

To eliminate the negative effects of existing methods such as social distance rule violation, slow test time and to create a pre-diagnosis method, deep learning sound analysis work has been carried out for Covid-19 disease, which turned into pandemic. For this purpose, experiments were performed on crowdsourced Coswara dataset Detection from Cough, Breath Speech Sounds with Short-Time Fourier Transform CNN Model. On dataset, tested samples selected model was trained different type sounds. The best result achieved cough-heavy 0.980 precision, 0.998 AUC, 0.990 F1-score set.

Language: Английский

Citations

0
A Comparative Study of Hybrid Deep Learning Techniques for COVID-19 Detection based on Cough Sound Analysis DOI

Ramya Polaki,

R Annamalai

2022 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS), Journal Year: 2023, Volume and Issue: unknown, P. 478 - 485

Published: Nov. 3, 2023

The COVID-19 pandemic, produced by the SARS-CoV-2 virus, has severe global consequences, resulting in substantial loss of life and posing a serious threat globally. Cough is common sign consequence COVID-19. sound analysis ability to help determine an individual's status. Using deep learning models, this study aims improve accuracy identification based on cough sounds. work employs twelve separate deep-learning models that were extensively trained COUGHVID dataset such as CNN, LSTM, BiLSTM, CNN-LSTM, CNN-BiLSTM with SGD, Adamax optimizer, Attention-based CNN-LSTM Adamax, SGD RMSProp optimizer. To overcome class imbalance, procedures pitch shifting time-frequency masking are used increase positive class. Among these variants, integration attention mechanism model convolutional neural network (CNN), bidirectional long short-term memory (Bi-LSTM) optimizer achieved highest validation accuracy, reaching 95.34%, precision 94.40 %, recall 95.50%, Fl-score 94.44%.

Language: Английский

Citations

0