Development and Applications of a Novel Acoustic Model for Respiratory System

呼吸是由高度复杂和动态的系统控制的。然而，目前还没有方便、非侵入性的技术来评估呼吸系统的动态变化。呼吸音是由空气通过咽部和其余呼吸道产生的，但目前还没有简单而动态的咽部声学模型来研究呼吸声产生的机制，以及如何使用呼吸声分析来监测咽部解剖和生理的变化。此外，与呼吸有关的数据，如呼吸率、呼吸强度和呼吸声的频谱特征，可以代表呼吸系统的特征。然而，目前可用于分析呼吸信号的算法在很大程度上局限于检查呼吸频率，这并没有结合呼吸信号产生过程中的动态变化。因此，有必要开发生理学驱动的数据挖掘算法来分析呼吸信号，以检查呼吸系统及其随时间的动态变化。*这个发现资助项目的重点将是首先开发新的声学算法来评估呼吸系统动力学，并培训HQP具有信号处理、物理、流体动力学和生理学的专门知识。AIM 1本身是一种新的方法-它将产生第一个使用呼吸声音来估计咽部狭窄、咽部潜在塌陷位置以及咽部组织成分和僵硬的模型。通过这种方法获得的知识可以用于睡眠呼吸暂停患者的表型，预测每个患者的最佳治疗方案，并制定客观和针对患者的措施来评估不同治疗方法的疗效。目标2将开发第一个生理学驱动的数据挖掘算法，以监测、预测和潜在地预防呼吸系统的长期异常。呼吸系统的异常会对个人的健康、社会/行为表现和生活质量产生短期和长期的影响。例如，睡眠中的呼吸障碍与与汽车和工作相关的事故、抑郁、白天过度嗜睡和在学校表现不佳的风险增加有关。*虽然这项提案的具体目标是基于睡眠呼吸暂停，但这些算法可以在呼吸信号的广泛应用中使用。这种应用可能包括监测呼吸能力以调整运动员的表现，防止工作场所因疲劳或嗜睡而造成的危险和伤害，防止因疲劳而发生的机动车碰撞，以及将歌唱过程中咽喉的动态变化建模为训练歌手和避免声带损伤的生物反馈。**

Breathing is controlled by a highly complex and dynamic system. However, there is no convenient and non-invasive technology to assess the dynamic changes in the respiratory system. Respiratory sounds are generated by the passage of air through pharynx and the rest of airways, but there is no simple and dynamic acoustic model of the pharynx in order to investigate the mechanisms of respiratory sound generation and how respiratory sounds analysis can be used to monitor variations in the pharyngeal anatomy and physiology. Furthermore, respiratory-related data such as breathing rate, intensity and spectral features of breathing sounds can represent characteristics of the respiratory system. However, currently available algorithms for analyzing respiratory signals are largely limited to examining respiratory rate, which does not incorporate the dynamic changes in the generation of respiratory signals. Therefore, there is a need to develop physiology-driven data mining algorithms for analysing respiratory signals to examine the respiratory system and its dynamic changes over time.***The focus of this Discovery Grant project will be first to develop novel acoustic algorithms to assess respiratory system dynamics and to train HQP with a specific knowledge of signal processing, physics, fluid dynamics, and physiology. Aim 1 is a novel approach in itself- it will yield the first model that uses respiratory sounds to estimate pharyngeal narrowing, potential sites of collapse in the pharynx, and pharyngeal tissue composition and stiffness. The knowledge gained by this method can be used to phenotype patients with sleep apnea, to predict the optimum treatment for every patient, and to develop objective and patient-specific measures to assess the efficacy of different treatments for sleep apnea. Aim 2 will develop the first physiology-driven data mining algorithm to monitor, predict, and potentially prevent long-term abnormalities of the respiratory system. Abnormalities in the respiratory system will have both short-term and long-term effects on the individual's health, social/behavioral performance, and quality of life. For example, breathing disorders during sleep are associated with increased risk of car and work-related accidents, depression, excessive daytime sleepiness, and poor performance at school. ***While the specific aims of this proposal are based on sleep apnea, these algorithms can be used in a wide range of applications for respiratory signals. Such application may include monitoring respiratory capacity to modify performance of athletes, preventing work place risks and injuries due to fatigue or sleepiness, preventing motor-vehicle collisions due to fatigue, and modeling the dynamic changes in the pharynx and larynx during singing as a biofeedback to train singers and to avoid vocal fold damage.**