OSAHS Prediction Research Based On Multi-mode Data Fusion Of Internet Of Things

Obstructive Sleep Apnea Syndrome(OSAHS)is mainly manifested in the pathological structure of upper trachea collapse or obstruction,often accompanied by snoring,sleep apnea,etc.Under extreme conditions,it may cause insufficient breathing at night and cause cardiovascular disease.death.At present,the medical diagnostic standard polysomnography(PSG)and most screening tools such as sleep pillows,ventilators,etc.have two defects: on the one hand,the operation is complicated and costly,and cannot be widely used in multiple scenarios;on the other hand "Contact" monitoring seriously interferes with the user’s sleep and affects the accuracy of OSAHS diagnosis.Therefore,this paper studies OSAHS related prediction technologies and methods,proposes a multi-modal data fusion OSAHS prediction system scheme based on the Internet of Things,designs and implements a snoring detection system,OSAHS prediction model and system application platform.The main content of this article And innovations include:(1)Aiming at the snoring detection system,using an algorithm based on the combination of multi-window spectral subtraction and short-term improved sub-band spectral entropy method to detect and correct snoring,proposed the suspected apnea index(SAHI)as the OSAHS feature expression of snoring,and designed Based on the Raspberry Pi’s snoring data acquisition and detection system,experiments show that the signal-to-noise ratio is improved by 12.39 d B,and the SAHI value detected by severe patients is compared with the gold standard PSG respiratory disorder index(AHI)value with an accuracy of 84.6%.(2)Aiming at the OSAHS prediction model,combining SAHI value,snoring information and age,body mass index(BMI),gender and other OSAHS highly correlated risk factors,an OSAHS prediction model based on multimodal data fusion is proposed and BP neural network tools are used Trying to build it,the results show that the effect of the prediction model is far better than that based on a single mode,with comprehensive sensitivity and specificity reaching82.6% and 77.1% respectively.(3)Aiming at the system application platform,we propose an integrated solution for detecting applications based on the Internet of Things through Raspberry Pi,cloud services,and We Chat applets.The snoring detection module,model deployment,database,and applet terminal are separately implemented.The design and implementation have achieved the expected results after comprehensive evaluation of the system,verifying the feasibility and reliability of the system.After testing and verification,the proposed scheme has the characteristics of non-contact,convenient operation and wide application in multiple scenarios,which can effectively assist OSAHS diagnosis and provide a certain reference value for OSAHS prediction research.