Advancements in AI-Guided Analysis of Cough Sounds for COVID-19 Screening: A Comprehensive Review

Abstract

The COVID-19 pandemic, a severe lung condition that produces excessive coughing, has placed an undue strain on healthcare systems throughout the globe. Screening that is dependable, inexpensive, and simple to get becomes critical for COVID-19. Even if the symptoms of severe and moderate diseases vary, coughing is still regarded as one of the most crucial indicators. Tools led by artificial intelligence (AI) have made it feasible to discover and test COVID-19 infections by using cough sounds for mass screening in areas with limited resources. In this paper, we examine the state of the art from 2020 to 2022 by examining AI-guided tools that employ machine learning and deep learning algorithms to analyze cough sounds for COVID-19 screening. We utilized the terms "(Cough OR Cough Sounds OR Speech) AND (Machine Learning Deep Learning OR Artificial Intelligence) AND (COVID-19 OR coronavirus)" in our research. To execute a better meta-analysis, we sought for a suitable dataset (size and source), algorithmic components, and performance ratings. We avoided pre-prints since they are not peer-reviewed; however, we did include publications from IEEE Explore, PubMed, Science Direct, ARXIV, and Springer Link to ensure we didn't overlook any current studies based on experimental research. This might be utilized for both early and long-distance diagnosis, assisting the globe in its battle against the epidemic. However, the best state-of-the-art approach for screening for COVID-19 was examined in this research.