http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
장수빈(Subin Jang),강윤호(Yunho Kang),박성민(Sungmin Park) 대한인간공학회 2019 대한인간공학회 학술대회논문집 Vol.2019 No.5
The purpose of this study is to develop a convolutional neural networks (CNN) based model that can diagnose pertussis using cough sound. Pertussis is a highly contagious respiratory disease, and 75% of the infected patients are infants less than 9 years old. In particular, if a child suffers from pertussis, it can die from complications and newborns within 4 weeks of birth have a 4% mortality rate due to pertussis. Early diagnosis is important to prevent the contagion and death from pertussis. Currently the method of diagnosing pertussis is performed by clinical symptoms and laboratory tests. However, this method is costly and difficult to diagnose it early. This study suggests a learning model for early diagnosis of pertussis. The cough sound data are transformed to image using MFCC and the hidden features are learned using the CNN model. The accuracy of the model has been improved through data augmentation and model optimization. The suggested model has high diagnostic performance: accuracy of 93%, sensitivity of 91%, and specificity of 80%. The proposed model with high classification accuracy and usability can be applied to medical applications to detect pertussis and it can be useful for early screening of children"s pertussis.