Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Developing a Multi-variate Prediction Model For COVID-19 From Crowd-sourced Respiratory Voice Data

Yuyang Yan, Wafaa Aljbawi, Sami O. Simons, Visara Urovi

TL;DR

This study investigates non-invasive detection of COVID-19 from crowd-sourced speech data using a range of machine learning and deep learning models. It evaluates traditional classifiers (LR, SVM), CNN, LSTM, and the end-to-end HuBERT model on the Cambridge COVID-19 Sound database, with HuBERT achieving the best performance (AUC 0.93, accuracy 0.86). External validation on the Coswara dataset confirms generalization (HuBERT AUC 0.83; accuracy 0.82), and the models can distinguish COVID-19 from cold symptoms with AUC up to 0.90. The findings suggest a low-cost, scalable, non-invasive screening approach using speech alone, with potential applicability in resource-limited settings and future exploration of interpretability and clinical integration.

Abstract

COVID-19 has affected more than 223 countries worldwide and in the Post-COVID Era, there is a pressing need for non-invasive, low-cost, and highly scalable solutions to detect COVID-19. We develop a deep learning model to identify COVID-19 from voice recording data. The novelty of this work is in the development of deep learning models for COVID-19 identification from only voice recordings. We use the Cambridge COVID-19 Sound database which contains 893 speech samples, crowd-sourced from 4352 participants via a COVID-19 Sounds app. Voice features including Mel-spectrograms and Mel-frequency cepstral coefficients (MFCC) and CNN Encoder features are extracted. Based on the voice data, we develop deep learning classification models to detect COVID-19 cases. These models include Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) and Hidden-Unit BERT (HuBERT). We compare their predictive power to baseline machine learning models. HuBERT achieves the highest accuracy of 86\% and the highest AUC of 0.93. The results achieved with the proposed models suggest promising results in COVID-19 diagnosis from voice recordings when compared to the results obtained from the state-of-the-art.

Developing a Multi-variate Prediction Model For COVID-19 From Crowd-sourced Respiratory Voice Data

TL;DR

This study investigates non-invasive detection of COVID-19 from crowd-sourced speech data using a range of machine learning and deep learning models. It evaluates traditional classifiers (LR, SVM), CNN, LSTM, and the end-to-end HuBERT model on the Cambridge COVID-19 Sound database, with HuBERT achieving the best performance (AUC 0.93, accuracy 0.86). External validation on the Coswara dataset confirms generalization (HuBERT AUC 0.83; accuracy 0.82), and the models can distinguish COVID-19 from cold symptoms with AUC up to 0.90. The findings suggest a low-cost, scalable, non-invasive screening approach using speech alone, with potential applicability in resource-limited settings and future exploration of interpretability and clinical integration.

Abstract

COVID-19 has affected more than 223 countries worldwide and in the Post-COVID Era, there is a pressing need for non-invasive, low-cost, and highly scalable solutions to detect COVID-19. We develop a deep learning model to identify COVID-19 from voice recording data. The novelty of this work is in the development of deep learning models for COVID-19 identification from only voice recordings. We use the Cambridge COVID-19 Sound database which contains 893 speech samples, crowd-sourced from 4352 participants via a COVID-19 Sounds app. Voice features including Mel-spectrograms and Mel-frequency cepstral coefficients (MFCC) and CNN Encoder features are extracted. Based on the voice data, we develop deep learning classification models to detect COVID-19 cases. These models include Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) and Hidden-Unit BERT (HuBERT). We compare their predictive power to baseline machine learning models. HuBERT achieves the highest accuracy of 86\% and the highest AUC of 0.93. The results achieved with the proposed models suggest promising results in COVID-19 diagnosis from voice recordings when compared to the results obtained from the state-of-the-art.
Paper Structure (19 sections, 5 figures, 3 tables)

This paper contains 19 sections, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Introduction
  3. Background
  4. Related work
  5. Methods
  6. Dataset Description
  7. Models
  8. Traditional Machine Learning Classifiers
  9. Convolutional Neural Network (CNN)
  10. Long Short-Term Memory (LSTM)
  11. Hidden-Unit BERT (HuBERT)
  12. Feature extraction
  13. Results
  14. Performance Measurements
  15. Experimental Results with different models
  16. ...and 4 more sections

Figures (5)

  • Figure 1: The used pipeline for both traditional Machine learning classifiers and Deep Learning classifiers for COVID-19 binary classification (e.g., COVID-19 vs. non-COVID-19).
  • Figure 2: Users characteristics (a) age, (b) gender, (c) COVID-19 test results, (d) the number of admissions to hospital
  • Figure 3: ROC curve for Models
  • Figure 4: ROC curve for Coswara dataset validation
  • Figure 5: ROC curve for distinguishing COVID-19 from cold symptoms