An adaptive speech signal processing for COVID-19 detection using deep learning approach.

Area of Science:

• Artificial Intelligence
• Medical Diagnostics
• Signal Processing

Background:

• The COVID-19 pandemic necessitates rapid and accessible diagnostic tools.
• Current reverse transcription polymerase chain reaction (RT-PCR) testing is costly, time-consuming, and can compromise social distancing.
• There is a need for innovative methods to detect COVID-19 efficiently.

Purpose of the Study:

• To develop a deep learning model for rapid COVID-19 detection using speech signals.
• To evaluate the effectiveness of generative adversarial networks (GANs) in classifying COVID-19 indicators in voice data.
• To compare the proposed GAN model's performance against existing neural network architectures.

Main Methods:

• A two-stage system involving signal pre-processing and classification was implemented.
• The least mean square (LMS) filter was employed for noise reduction in speech signals.
• Mel-frequency cepstral coefficients (MFCCs) were extracted and analyzed using a GAN for classification.

Main Results:

• The generative adversarial network (GAN) model demonstrated high accuracy in distinguishing between COVID-19 and non-COVID-19 speech signals.
• Mel-frequency cepstral coefficients (MFCCs) showed a strong correlation with COVID-19 related cough and breathing sounds.
• The proposed GAN achieved superior performance compared to Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN).

Conclusions:

• Deep learning, specifically GANs, offers a promising avenue for non-invasive, rapid COVID-19 detection via speech analysis.
• The developed system provides high precision (96.54%), recall (96.15%), and accuracy (98.56%).
• This approach presents a viable, cost-effective, and efficient alternative to traditional RT-PCR testing.