Personalized fitting and evaluation of hearing aids with EEG responses

据估计，英国约有600万人可以从助听器中受益，但只有大约200万助听器使用者，其中只有70%经常使用助听器。现代助听器是复杂的设备，具有先进的功能（不同频段的增益，幅度压缩，反馈消除，降噪，定向麦克风等），需要专业人员来安装。助听器的益处有限是许多患者不经常使用助听器的主要原因。传统上，助听器的安装主要基于“听力图”，它告诉患者在不同频率下可以听到的最安静的声音（短音调），并从患者自愿和主观的反应（通常通过点击按钮）中获得逐渐安静的声音。然而，很明显，听图只能提供对听力损失的有限了解，仅基于此来安装助听器可能会导致对患者最重要的结果非常不同，即理解语言。在噪音中理解言语的困难是助听器使用者的主要抱怨之一。目前的项目旨在为每个患者改善个性化的助听器。关键技术将是通过分析从放置在头皮上的电极获得的脑电图（EEG）反应，直接测量大脑对声音的反应。这种分析是“客观的”，不需要患者对刺激做出自愿的和“主观的”（并不总是可靠的）反应。我们认为这很重要，因为它可以在无法提供这种自愿反应的患者中进行，例如婴儿或老年痴呆症患者。通过不间断地监测听力来评估患者的感知，助听器的性能也可以在自然听力条件下和更长的时间内进行评估。最终，随着用户的需求和听觉环境的变化，这种方法也可能允许在没有听力学家在场的情况下调整助听器的设置。我们将为该项目开发的测试刺激（听力挑战）将包括比目前在诊所常规使用的更广泛的声音，允许对听力损失进行更细微的（鉴别）诊断，并关注对言语（包括噪音中的言语）的反应。该项目的主要研究目标是通过计算机分析脑电图对复杂现实世界信号的反应，实现对大脑（从耳蜗到脑干和大脑皮层）中听力功能和语言处理的可靠评估。这提出了重大的科学和技术挑战，需要开发新的语音和脑电图数据信号分析方法，这可能与听力障碍、认知以及助听器设置和性能有关。这些重大挑战和对患者利益的关注的结合使这成为一个令人兴奋和冒险的项目。本建议书的主要目标是提出、评估和建议：从脑电信号中提取听力表现和患者言语获取信息的信号处理方法2。用于评估听力的刺激物这项跨学科的工作将在大学（听力科学、语音处理、信号分析）、行业（听力技术）和患者（选择听力挑战）之间进行合作。开展这项工作预计将惠及患者及其家属和护理人员（通过使用助听器提高生活质量）、保健服务（提高效率）、工业（新的诊断技术）和科学界（更好地了解听力；改进分析脑电图信号的方法）。

It has been estimated that some 6 million people in the UK could benefit from hearing aids, but there are only approximately 2 million hearing aid users, and of these, only 70% use their hearing aids regularly. Modern hearing aids are complex devices with advanced features (gain in different frequency bands, amplitude compression, feedback cancellation, noise reduction, directional microphones etc.) and require professionals to fit them. Limited benefit from hearing aids is a major reason why many patients do not use their devices regularly. Conventionally, hearing aids are fitted based primarily on the 'audiogram', which informs on the quietest sounds (short tones) that the patient can hear at different frequencies and is obtained from patients' voluntary and subjective response (usually by clicking a button) to progressively quieter sounds. However, it is clear that the audiogram only provides limited insight into hearing loss, and fitting hearing aids based on this alone can lead to very diverse results in what is of most importance to patients, namely understanding speech. The difficulty of understanding speech in noise is one of the chief complaints of hearing aid users. The current project aims to improve personalized fitting of hearing aids to individual patients. The key technique will be the use of measurements taken directly from the brain's response to sound, by analysing the electroencephalographic (EEG) responses obtained from electrodes placed on the scalp. The analysis is 'objective', without requiring patients' voluntary and 'subjective' (and not always reliable) response to stimuli. We think this is important as it can be carried out in patients who are unable to provide such voluntary responses, for example infants or the elderly with dementia. By monitoring hearing without constant interruption to assess patients' perception, the performance of the hearing aid can also be assessed in natural listening conditions and over a longer time period. Ultimately this approach may also allow hearing aid settings to be adjusted without the presence of an audiologist, as users' needs and the auditory environment change. The test stimuli (hearing challenges) we will develop for the project will include a wider range of sounds than are currently routinely used in clinics, allowing for more subtle (differential) diagnosis of hearing loss, and a focus on the response to speech (including speech-in-noise).The key research aim in this project is to achieve a robust assessment of hearing function and speech processing in the brain (from the cochlea to the brain stem and cerebral cortex) by the computer analysis of EEG responses to complex real-world signals. This presents major scientific and technical challenges, needing the development of novel signal-analysis methods for speech and EEG data, which can be related to hearing impairment, cognition, as well as hearing aid settings and performance. The combination of these major challenges and a focus on patient benefit makes this an exciting and adventurous project. The main objectives of this proposal are to propose, assess and recommend:1. Signal processing methods to extract information from EEG signals on hearing performance and patients' access to speech2. Stimuli to use in assessing hearing3. Algorithms to optimize hearing aid fitting, based on parameters extracted from EEG responses This interdisciplinary work will be carried out as a collaboration between universities (hearing science, speech processing, signal analysis), industry (hearing technologies) and patients (choosing hearing challenges). The benefits of undertaking this work are expected to be to patients and their family and carers (improved quality of life from using hearing aids), the health-services (improved efficiency), industry (new diagnostic technologies) and the scientific community (better understanding of hearing; improved methods for analysing EEG signals).