Coronavirus diagnosis using cough sounds: Artificial intelligence approaches.

Area of Science:

• Medical Informatics
• Artificial Intelligence
• Data Mining

Background:

• The COVID-19 pandemic necessitated rapid identification of infected individuals for containment.
• Artificial intelligence and data mining offer potential for cost-effective disease prevention and management.

Purpose of the Study:

• To develop data mining models for diagnosing COVID-19 using cough sounds.
• To explore the efficacy of various machine learning algorithms for cough-based COVID-19 detection.

Main Methods:

• Supervised learning classification algorithms were employed, including Support Vector Machine (SVM), random forest, and Artificial Neural Networks (ANNs).
• Specific ANNs utilized were Fully Connected, Convolutional Neural Networks (CNNs), and Long Short-Term Memory (LSTM) recurrent networks.
• Data comprised approximately 40,000 cough recordings collected during the COVID-19 pandemic.

Main Results:

• The developed models demonstrated acceptable accuracy in diagnosing COVID-19 from cough sounds.
• An average accuracy of 83% was achieved across models.
• The best-performing model reached an accuracy of 95%.

Conclusions:

• The findings support the reliability of using cough sounds with AI for COVID-19 screening and early diagnosis.
• This method shows promise for developing accessible diagnostic tools.
• Even simpler AI networks can yield acceptable results, indicating broad applicability.