Cough Sounds in Screening and Diagnostics: A Scoping Review

Area of Science:

• Biomedical Engineering
• Health Informatics
• Acoustic Signal Processing

Background:

• Cough sound analysis for diagnostics has a long history but has seen accelerated academic interest recently.
• The potential for cough sounds as a low-cost, noninvasive disease biomarker is significant.
• Ubiquitous mobile technology offers high-quality recording capabilities for cough analysis.

Purpose of the Study:

• To review applications of cough sounds in biomedical and engineering literature for screening and diagnostics.
• To focus on disease types, data collection, processing, analytics, accuracy, and limitations.
• To synthesize current research trends and identify research gaps.

Main Methods:

• A comprehensive scoping review of multiple databases (PubMed, EMBASE, Scopus, etc.) was performed.
• Searched literature from inception to August 2021, including peer-reviewed and gray literature.
• Included studies on screening and diagnostic uses of cough sounds in humans and animals.

Main Results:

• 108 articles met inclusion criteria from 438 abstracts screened.
• Human studies dominated (77.8%), primarily focusing on adults (57.3%).
• Machine learning was used in 61.1% of studies, with most published after 2010. Common focuses included cough detection (41.7%) and COVID-19 screening (11.1%). Asthma diagnosis was prevalent in pediatric studies (52.6%).

Conclusions:

• Cough sound analysis is a rapidly growing field, driven by machine learning and mobile technology.
• Nonstandardized data collection protocols and limited dataset validity remain significant challenges.
• Further research with standardized multimodal data collection is needed to improve diagnostic accuracy and external validity.