Detection And Identification Of Cough By Mattress-style And Belt-style Physiologic Signal Monitoring Systems

In recent years, the mist appears in China frequently, so air pollution has become more and more critical. The respiratory system disease which initiated by terrible air has been one of the problems threatening our health. Cough is the common and typical symptom of respiratory system disease. What’s more, frequent and severe cough can make patients feel painful in the chest and bring a bad influence on their lives. So, how to make doctors diagnose the disease precisely by effectively quantifying the frequency and intensity of cough has been the focus on cough research recently.Nowadays, the cough quantification is mainly long-term monitoring patients by cough monitor system, and then analyzing automatically the frequency and intensity of cough in identifying algorithm. In general, cough monitoring method is to combine voice recording with voice identifying technology. This paper proposes a new method, analyzing the movements and respiratory signal on the belly from many sides which is based on mattress-style and belt-style physiologic signal monitoring systems respectively, and identifying the cough successfully. Mattress-style monitoring system’s essential part is a micro-movement sensitive inflatable mattress based on pressure sensors without sticking any electrode on bodies. Patients can be measured all kinds of physiologic signals lay in mattress without wearing any instrument. Belt-style monitoring system consists of a fixing belt and electronic parts inside the belt, such as acceleration sensor, microphone, and pressure sensors and so on. It records acceleration on the belly, vibration and respiratory waves, and obtains envelopes of sounds by processing original voice signals. The main tasks of this paper(1) Pre-processing the original signals obtained in mattress-style system and extracting chest impact, respiration waves and leg impact from them successfully. In terms of the frequency domain in obtained signals, this paper combines digital signal filtering with wave analysis to process signals. Incorporated the fundamentals of digital filters with practice demands, zero-phase filter is chosen and designed, and obtains zero-phase distortion output signals. Then, wave analysis is resolving the original signals to have different frequency domain information and choose ones up to the requirements. These two methods establish good foundation for further research.(2) Extracting envelopes of sound signals from belt-style system. Before processing original sound signals, having the power spectrum in different patients’ cough samples and ensuring cough frequency range which is the frequency of high-pass filter. Then, sound signals keep high frequency components through high-pass filter. Last, summarize and lack-sampling high frequency components to get envelope waves. The advantages of adopting envelopes are less data, neat waves and highlighting cough which favors writing algorithm.(3) In view of two different systems, this paper designs experiments respectively. Recording patients’voice signals all day long through microphone is used for manual counting cough. About experiment data, this paper develops algorithms matching systems to identify cough. The cough can be distinguished from interference (clearing throat, talking and so on) splendidly by making full use of unique features of acceleration, respiration waves, vibration and sound envelopes. Compared to manual counting, the sensitiveness and specificity of the algorithm can reach to88.4%-90.0%and96.0%～98.6%respectively. Mattress-style and belt-style monitoring systems have completed monitoring and recognizing cough preliminarily.In the future, we can in-depth study the change in every physiologic signal when cough happens, and classify automatically different types of cough. Moreover, enhancing the sensitivity and specificity of cough monitoring system and decreasing the interference which the surroundings bring make it possible to apply in clinical diagnosis.