Identifying COVID-19 by using spectral analysis of cough recordings: a distinctive classification study

Area of Science:

• Medical Diagnostics
• Bioacoustics
• Computational Biology

Background:

• Early diagnosis of respiratory diseases, including COVID-19, is crucial for effective treatment and preventing irreversible health effects.
• The COVID-19 pandemic highlighted the need for rapid and accurate diagnostic tools, especially for differentiating symptoms from similar illnesses like influenza.
• Cough sound analysis has emerged as a promising non-invasive method for disease detection.

Purpose of the Study:

• To investigate the potential of using cough sound analysis for the specific detection and classification of COVID-19.
• To evaluate the efficacy of Mel-frequency cepstral coefficients (MFCC) and Support Vector Machine (SVM) algorithms in distinguishing COVID-19 coughs from other coughs.
• To establish a reliable and accurate method for identifying COVID-19 based on acoustic features of coughs.

Main Methods:

• A dataset of cough sounds was collected from 16 subjects, categorized into COVID-19 positive and negative groups via polymerase chain reaction (PCR) testing.
• Feature extraction was performed using Power Spectral Density (PSD) via Short-Time Fourier Transform (STFT) and Mel-frequency cepstral coefficients (MFCC).
• A Support Vector Machine (SVM) classifier, utilizing a Radial Basis Function (RBF) kernel, was employed to classify the cough sounds.

Main Results:

• The SVM algorithm, combined with MFCC feature extraction, achieved a classification accuracy of 95.86% in identifying COVID-19 coughs.
• The diagnostic performance demonstrated high sensitivity (98.6%) and specificity (91.7%) for detecting COVID-19 based on cough sounds.
• Signal frequency spectrum analysis, particularly 3D plots, aided in revealing salient features for cough differentiation.

Conclusions:

• Cough sound analysis using MFCC and SVM presents a highly accurate and effective method for COVID-19 diagnosis.
• This non-invasive technique offers a valuable tool for early detection and differentiation of COVID-19, complementing traditional diagnostic methods.
• The study underscores the potential of bioacoustics in public health surveillance and disease diagnostics, particularly during pandemics.