Development and technical validation of a smartphone‐based pediatric cough detection algorithm

Abstract

Introduction:Coughing is a common symptom in pediatric lung disease and cough frequency has been shown to be correlated to disease activity in several conditions. Automated cough detection could provide a noninvasive digital biomarker for pediatric clinical trials or care. The aim of this study was to develop a smartphonebased algorithm that objectively and automatically counts cough sounds of children. Methods:The training set was composed of 3228 pediatric cough sounds and 480,780 noncough sounds from various publicly available sources and continuous sound recordings of 7 patients admitted due to respiratory disease. A Gradient Boost Classifier was fitted on the training data, which was subsequently validated on recordings from 14 additional patients aged 0-14 admitted to the pediatric ward due to respiratory disease. The robustness of the algorithm was investigated by repeatedly classifying a recording with the smartphone-based algorithm during various conditions. Results:The final algorithm obtained an accuracy of 99.7%, sensitivity of 47.6%, specificity of 99.96%, positive predictive value of 82.2% and negative predictive value 99.8% in the validation dataset. The correlation coefficient between manualand automated cough counts in the validation dataset was 0.97 (p < .001). The intra-and interdevice reliability of the algorithm was adequate, and the algorithm performed best at an unobstructed distance of 0.5-1 m from the audio source. Conclusion:This novel smartphone-based pediatric cough detection application can be used for longitudinal follow-up in clinical care or as digital endpoint in clinical trials.

Respiratory Diseases Diagnosis Using Audio Analysis and Artificial Intelligence: A Systematic Review

“…The first category pertains to the detection of cough sounds, involving the identification of a sound signal as a cough. Within this category, various authors [ 7 , 14 , 17 , 18 , 21 , 22 , 23 , 24 , 25 , 28 , 29 ] have achieved detection accuracies as high as 99.64%.…”

“…During the data preparation stage, a challenge that most studies attempt to tackle is the isolation of cough events from the raw signal. Studies suggest a variety of methods for labeling the raw signals such as utilizing a moving window signal deviation as a function of time [ 8 ] and splitting the signal based on the silence segments [ 15 ], manual labeling, by using software such as PRAAT [ 17 ] and others [ 31 ], or even empirically [ 4 , 7 , 10 , 16 , 18 , 20 , 23 , 26 , 28 , 29 , 30 , 36 ]. Among other methods are classifiers that are able to distinguish the energy levels among different segments of a signal [ 12 ] and finally, Empirical Mode Decomposition (EMD) [ 11 ].…”

“…In Refs. [ 18 , 26 ], popular software (e.g., openSMILE [ 42 ]) is utilized to extract a diverse set of features, which are subsequently subjected to feature selection techniques to retain only the most significant ones for the final feature matrix. Conversely [ 29 ], primarily employs Fast Fourier Transform (FFT) to extract features from the cough signals.…”

Cough classification of MEMS microphone signal using time-series and tabular machine learning algorithms

“…In the former case, the observational data samples are assumed to be independent and identically distributed (i.i.d.). 7–27 However, in the latter case, the order of the data samples carries intrinsic information. These tabular methods themselves show excellent performances in various fields of application such as healthcare and medical predictions.…”

“…4 Many studies have attempted automatic cough detection. [5][6][7][8][9][10][11][12][13][14][15] With regard to the sensor type, the majority of the studies have used audio sensors, such as audio microphones or contact microphones. 5,6 Birring et al 5 collected cough signals from patients using an audio acquisition system called the Leicester Cough Monitor.…”