In this work, we give a short introduction on cough detection efforts that were undertaken during the last decade and we describe the solution for automatic cough detection developed for the AioCare portable spirometry system. As the system is intended to be used in a large variety of environments and different patients, we train the algorithm using the large database of spirometry curves which is the NHANES database by the American National Center for Health Statistics. We apply few data preprocessing steps and train different classifiers such as logistic regression (LR), feed forward artificial neural network (ANN), artificial neural network combined with principal component analysis (PCA-ANN), support vector machine (SVM) and random forest (RF) on this data to choose the one of the best performance. The accuracy, sensitivity and specificity of the classifiers were comparable and equaled within the range 91.1{\div}91.2%, 81.8{\div}83.8% and 95.0{\div}95.9% for the test set, respectively. The ANN solution was selected as the final classifier. Classification methodology developed in this study is robust for detecting cough events during spirometry measurements. We also show that it is universal and transferable between different systems as the performance on the NHANES and AioCare test sets is similar. The solution presented in this work was implemented in the AioCare mobile spirometry system.

介绍了过去十年咳嗽检测的研究，描述了AioCare便携式肺活量检测系统的自动咳嗽检测解决方案，该解决方案基于气流信号而非音频信号，使用逻辑回归、神经网络、支持向量机等分类器进行训练并选择了人工神经网络，测试表明该分类器精度稳健，能够应用于不同的环境和患者，是第一个基于气流信号的自动咳嗽检测算法描述，并且是第一个应用于商用肺活量检测系统的咳嗽检测算法。

基于气流信号的便携式肺功能检测系统自动咳嗽检测