Wearable, Always-on Stethoscope for Early Detection of Asthma Attack

摘要 本研究旨在开发和测试一种可穿戴的，始终在线的听诊器，以提供解决方案， 对呼吸道症状量化的需求未得到满足。这样的设备是必要的准确诊断 在680万婴儿、幼儿和其他患有哮喘的人群中， 智力残疾，无法报告其呼吸道症状或进行肺功能测试 美国的通过准确的诊断和控制评估，可以开始适当的哮喘治疗 及时减少哮喘的不良后果所提出的可穿戴设备所需的关键要素 听诊器（即，没有大体积声学耦合器的无线听诊器）是（1）谐振麦克风阵列 具有超过100 - 800 Hz的前所未有的声音检测能力和（2）超低功率信号处理。我们提出 开发一系列基于高Q（品质因数）谐振的声学滤波麦克风， 麦克风膜片，用于准确检测异常肺音。一组Q滤波和Q增强的 提出了一种麦克风，使得（1）可以避免特征提取滤波器和（2）非常小的肺 可以在没有声学耦合器的情况下检测来自胸部的声音。 肺音是非常困难的检测从胸部没有一个笨重的声学耦合器，因为声音 在自由空间中，在100 - 800 Hz的频率范围内，声压级（SPL）仅为22 - 30 dB SPL。这种 商用微型麦克风无法准确检测到声压级。因此，我们将开发和使用 Q为40 - 60的8个谐振麦克风阵列（采用微加工工艺制造），用于检测肺部 声音低至22 dB SPL，比人耳可以检测到的最低声音低4 dB， 将声音的频率分量分割成8个不同的窄带通过频率区域， - 800赫兹。 我们假设，持续动态监测肺音的声学特征， 哮喘将提高诊断的准确性和治疗儿童哮喘患者。目前，哮喘是 主要通过看护者病史和简短的诊室检查作为直接病史在幼儿中诊断 及呼吸功能测试。所提出的装置是完全新颖的 因为目前没有能够提供对哮喘征兆的全天候监测的设备。而 传统的麦克风可以检测咳嗽和明显的喘息，它们的实用性受到灵敏度不足的限制， 并且不适于连续的、流动的监测。 拟议的谐振麦克风阵列将与超低功耗电子器件集成，用于可穿戴设备。 一种听诊器，持续跟踪肺部声音，用于检测哮喘体征和症状，如咳嗽 和喘息，可能发生在运动或睡眠期间（当护理人员不在场时）， 被孩子们错误地报道或忽视。无线传输的异常肺音将通过 互联网拟议中的听诊器将为那些无法提供 准确的历史或自我管理的困难，如前语言儿童和个人， 智力残疾，以避免严重的哮喘发病率。该设备可以检测出哮喘失控的迹象， 通过互联网通知父母、看护者和医疗专业人员。这样做，我们可以改善 通过更准确地跟踪环境和行为触发因素， 可以用来改善管理。 为了测试可穿戴听诊器的潜在临床实用性，我们将首先记录肺音， 常规电子听诊器，并将其注释为“正常”、“喘息”、“咳嗽”， 等由一个专家评审团评审。注释的声音将受到选择为并行的频谱滤波。 所提出的谐振麦克风阵列的频率响应。模式识别算法将 应用于这些声音文件，并用于确定可穿戴听诊器的识别准确性。 四套可穿戴听诊器将在研究的第18、30、42和54个月交付 期间，并将用于记录肺音和测试自动分类准确性，这将被比较 诊断为哮喘控制良好或控制不良的患者。 最后，可穿戴听诊器将在门诊哮喘儿童患者中进行持续测试， 30天的时间，用于可穿戴听诊器检测哮喘发作的预测能力。

ABSTRACT This research aims to develop and test a wearable, always-on stethoscope to provide a solution to the unmet need for the quantification of respiratory symptoms. Such a device is necessary for the accurate diagnosis of asthma and assessment of asthma control in 6.8 million infants, young children and other populations with intellectual disability who are unable to report their respiratory symptoms or perform lung function testing in the United States. With accurate diagnosis and assessment of control, appropriate asthma therapy can be initiated without delays to minimize adverse asthma outcomes. The key elements needed in the proposed wearable stethoscope (i.e., a wireless stethoscope without a bulky acoustic coupler) are (1) resonant microphone array with unprecedented sound detectability over 100 – 800 Hz and (2) ultra-low power signal processing. We propose to develop a bank of acoustically-filtering microphones that are based on a high Q (quality factor) resonance of a microphone diaphragm, for accurate detection of abnormal lung sounds. A bank of Q-filtered and Q-enhanced microphones is proposed so that (1) feature-extracting filters may be avoided and (2) extremely small lung sounds can be detected from the chest without an acoustic coupler. Lung sounds are very difficult to detect from the chest without a bulky acoustic coupler, as the sound pressure level (SPL) is only 22 – 30 dB SPL, in free space, over a frequency range of 100 – 800 Hz. This kind of SPL cannot be accurately detected by a commercial miniature microphone. Thus, we will develop and use an array of 8 resonant microphones with Q of 40 – 60 (fabricated with a microfabrication process) to detect lung sounds down to 22 dB SPL, 4 dB lower than the lowest sound a human ear can detect, and to automatically segment the sound’s frequency components into 8 different narrowly-band-passed frequency regions over 100 – 800 Hz. We hypothesize that continuous ambulatory monitoring of lung sounds for acoustic characteristics of asthma will improve the diagnostic accuracy and treatment in pediatric asthma patients. Currently, asthma is diagnosed in small children primarily through caregiver history and brief in-office exam as direct patient history and respiratory function testing are not available for these young patients. The proposed device is entirely novel as currently there is no device capable of providing round-the-clock monitoring for signs of asthma. While conventional microphones can detect cough and overt wheezing, their utility is limited by insufficient sensitivity, and are not amenable for continuous, ambulatory monitoring. The proposed resonant microphone array will be integrated with ultralow power electronics for a wearable stethoscope that continuously tracks lung sounds for the detection of asthma signs and symptoms such as cough and wheeze which may occur during exercise or sleep (when caregivers are not present) and may be misreported or ignored by young children. Wirelessly transmitted abnormal lung sounds will be accessed through the internet. The proposed stethoscope will provide an unprecedented means for those incapable of providing an accurate history or difficulty with self-management, such as pre-verbal children and individuals with intellectual disability, to avoid serious asthma morbidity. The device detects signs of uncontrolled asthma and notifies the parent, caregiver and medical professionals through the internet. In doing so, we may improve the management of asthma patients through more accurate tracking of environmental and behavioral triggers which can be used to improve management. To test the potential clinical utility of the wearable stethoscope, we will first record lung sounds with conventional electronic stethoscopes from pediatric patients, and annotate them as “normal,” “wheeze,” “cough,” etc. by a panel of expert reviewers. The annotated sounds will be subjected to spectral filtering chosen to parallel the frequency response of the proposed resonant microphone array. A pattern recognition algorithm will be applied to these sound files and be used to determine the recognition accuracy of the wearable stethoscopes. Four sets of the wearable stethoscopes will be delivered at the 18th, 30th, 42nd and 54th month of the research period, and will be used to record lung sounds and test automatic classification accuracy, which will be compared to patients diagnosed with well- or poorly-controlled asthma as determined by consensus diagnostic criteria. Finally, the wearable stethoscopes will be tested in ambulatory asthmatic pediatric patients continuously over a 30 day period, for the predictive ability of the wearable stethoscope in detecting asthma attacks.