JMIR Publications

ABSTRACT

Background:

Asthma is one of the most prevalent chronic respiratory diseases. Despite increased investment in treatment, little progress has been made in the early recognition and treatment of exacerbations over the last decade. Nocturnal cough monitoring may provide an opportunity to identify patients at risk for imminent exacerbations. Recently developed approaches enable smartphone-based cough monitoring. These approaches, however, have not undergone longitudinal overnight testing nor have they been specifically evaluated in the context of asthma. Also, the problem of distinguishing partner cough from patient cough when two or more people are sleeping in the same room in contact-free audio recordings remains unsolved.

Objective:

The objective of this study was to evaluate the automatic recognition and segmentation of nocturnal asthmatic cough and cough epochs in smartphone-based audio recordings collected in the field. We also aimed to distinguish partner cough from patient cough in contact-free audio recordings by assigning cough to sex.

Methods:

We used a convolutional neural network (CNN), which we developed in prior work, for cough recognition. We further used techniques (e.g., ensemble learning, mini-batch balancing, and thresholding) to address the imbalance in the dataset. We evaluated the classifier in a classification task and a segmentation task. The cough recognition served as the basis for the cough segmentation from continuous audio recordings. We compared automated cough and cough-epoch counts to humanly annotated cough and cough-epoch counts. We employed Gaussian mixture models (GMM) to build a classifier for the assignment of cough and cough-epoch signals to sex.

Results:

We recorded audio data from 94 adults with asthma (57% females, 43 years old on average) during 29 nights with smartphones lying next to the bed in their everyday environment. Out of 704,697 sounds, we identified 30,304 sounds as cough. A total of 26,166 coughs occurred without a 2-s pause between coughs, yielding 8,238 cough epochs. The ensemble classifier performed well with a Matthews Correlation Coefficient of 92% in a pure classification task and achieved comparable cough counts to human annotators in the segmentation of coughing overnight. Mean difference between automated and observer cough counts was -0.1 coughs. Mean difference between automated and observer cough-epoch counts was 0.24 cough epochs. The GMM cough-epoch-based sex assignment performed best yielding an accuracy of 83%.

Conclusions:

Our study showed longitudinal nocturnal cough and cough-epoch recognition from smartphone-based audio recordings in the everyday nights of adults with asthma. It contributes to the distinguishing of partner cough from patient cough in contact-free recordings by assigning cough and cough-epoch signals to the corresponding sex of the patient. This research represents a step towards enabling passive scalable cough monitoring for adults with asthma. Clinical Trial: Trial registration number NCT03635710.