Study On Monitoring And Health Analysis Of Sleep Breathing State Based On Acoustic Signal

Obstructive sleep apnea hypopnea syndrome(OSAHS)is a common sleep disorder.At present,polysomnography(PSG)is the "gold standard" for the diagnosis of the disease.However,the number of PSG equipment in hospital is limited,the cost of diagnosis is expensive and the means of monitoring is complicated.Portable diagnosis method has become an urgent problem.Among the various signals detected by PSG equipment,the acoustic signal snore of patients plays an important role in the diagnosis of the disease.In order to facilitate the diagnosis at home,snoring is only used as the screening basis,designs rapid diagnostic methods,and develops mobile app for detection.The specific work includes the following three aspects:(1)Research on recognition method of snoring in night sleep based on Transfer Learning:It is difficult to recognize snoring accurately because of the complexity of sleep background sound.In this paper,convolutional neural network is used to learn snoring feature representation automatically and transfer learning,then a snore recognition model was established by using multiple loss logistic regression algorithm.The train,ng is carried out in combination with public acoustic dataset ESC-50 and snoring data collected.And it realizes the detection and recognition of sleep snore in the complex background of family.(2)Research on OSAHS diagnosis method based on snoring:To improve the diagnostic accuracy of OSAHS based on snoring,the deep learning feature is extracted in this paper,based on the traditional acoustic feature calculation;Then,a multinomial loss logic regression algorithm is used to build the OSAHS qualitative diagnosis model based on the depth feature.And it has a high accuracy compared with the traditional model.In order to further evaluate the patient's condition,a OSAHS quantitative diagnosis model based on depth feature is established by using deep neural network.The experimental results show that the correlation between the experimental results of the model and the diagnostic results of the PSG device reaches 87%.It can effectively assess the patient's respiratory disturbance index to get more detailed assessment results.(3)Research and implementation of home sleeping disorder detection system based on Android:Smart sleep detection mobile phone APP for home environment.First,sleep sound collection and snoring recognition in family environment are realized,secondly,the OSAHS quantitative diagnosis model based on deep neural network is applied to APP.The subjects could conduct respiratory quality assessment and OSAHS screening at home.It has been recognized in practical applications.