Sleeping Respiration Status Monitoring And Health Analysis Based On Bone-conducted Microphone

Snoring is a common sleep phenomeno which is easily ignored. Actually it is theperformance of breathing weaken and a sign of health problems. People who snore areassociated with significant apnea or hypopnea phenomenon, for most people,occasional transient respiratory obstruction is very common, but if the obstruction orhypopnea is very frequent and the respiratory obstruction is in serious condition, thenthis will become a disease we called Sleep Apnea Hypopnea Syndrome (SAHS).For a long time, the most used and most authoritative diagnosis method for theSAHS is polysomnography (PSG), but this method did not make the most SAHSpatients get effective treatment due to its high cost, the limited of place and time.Therefore, the patients desire a convenient and effective means to monitor thesleeping health. In this context, the mobile health about snoring test appeared.In this paper, we have designed and completed a portable sleep health monitoringequipment. This equipment is mainly composed of two parts: one is aim to themonitoring of the sleep breath states through the analysis of the sounds signal datacollected by bone conduction microphone, another part is composed of gyroscope andaccelerometers modules which can achieve the judgment of sleeping position by thefusion of the data from two modules.Compared with the air-conducted microphone, the bone-conducted microphonewith the characteristics of noise resistant and which can capture the weaker signalsthan the breathing or snoring will monitor the snoring and respiration status duringsleeping more accurately in the future. Until now, there are no bone-conductedmicrophones available on the market due to the technical limitations. So we use ashock-sensor combined with a set of self-developed amplifier to record and monitorthe respiration status. We also determine the optimal breath sound collection locationthrough a series of experiments.According to the characteristics of sleeping breath, we divided the breath intothree states: normal breath, snoring, stop breathing and built their model. Theexperimental results show that, when use the time domain boundary detection andHTK, we can identify and monitor the breath states more accurately and effectively.