Exploring machine learning for audio-based respiratory condition screening: A concise review of databases, methods, and open issues

Abstract

Auscultation plays an important role in the clinic, and the research community has been exploring machine learning (ML) to enable remote and automatic auscultation for respiratory condition screening via sounds. To give the big picture of what is going on in this field, in this narrative review, we describe publicly available audio databases that can be used for experiments, illustrate the developed ML methods proposed to date, and flag some under-considered issues which still need attention. Compared to existing surveys on the topic, we cover the latest literature, especially those audio-based COVID-19 detection studies which have gained extensive attention in the last two years. This work can help to facilitate the application of artificial intelligence in the respiratory auscultation field.

Applied artificial intelligence in healthcare: Listening to the winds of change in a post-COVID-19 world

“…Xia et al . 23 summarized publicly available data sets annotated by respiratory experts and reviewed the latest machine learning methods used for respiratory screening during the Covid-19 pandemic. Scaboro et al .…”

“…Tanwar A et al 21 proposed an unsupervised method that leverages external clinical knowledge and contextualized word embeddings by ClinicalBERT for numerical reasoning in different phenotypic contexts. Jana et al 22 23 summarized publicly available data sets annotated by respiratory experts and reviewed the latest machine learning methods used for respiratory screening during the Covid-19 pandemic. Scaboro et al 24 compared some of the current systems for detecting adverse drug events using social media data and proposed strategies to increase the robustness of these systems.…”

Transfer Learning based Diagnosis and Analysis of Lung Sound Aberrations

“…In recent years, lung sound diagnosis and classification have attracted much research attention [1][2][3][4][5][6][7]. Breathing is so necessary that in 24 hours, an average human can breathe 25,000 times [8][9][10][11][12]. According to the World Health Organization (WHO), the COVID-19 epidemic on May 24, 2021, there have been 166,860,081 verified cases, with 3,459,996 deaths recorded [13].…”

BrachySound: Machine Learning Based Assessment of Respiratory Sounds in Dogs

“…This makes it challenging, even for veterinarians, to determine which dogs require a surgical intervention 5,21,26 Aligned with other disciplines [27][28][29][30][31] , dogs' behavioral analysis seems to move toward the adoption of modern computational and data-driven models such as machine learning and deep learning-based models [32][33][34][35] .Indeed, recent advances in machine learning have shown the potential for objective automated respiratory sound analysis to serve as a biometric tool for assessing respiratory function in humans 36,37 .In a similar vein, Oikarinen et al 38 utilized deep convolutional neural networks for the classification of marmoset vocalizations, highlighting the potential of such techniques in animal sound classification and source attribution. Xia et al 39 reviewed machine learning methods for audio-based respiratory condition screening, discussing also the use of machine learning for auscultation by the respiratory system. In the field of human medicine, machine learning algorithms have been developed for the detection of cardiac murmurs using digital stethoscope platforms, showcasing the versatility and potential of these techniques across various medical applications 40 .…”