German-Russian statistical audiology: Data conditioning and audiological profile analysis for the diagnosis and compensation of hearing impairment” (GRUSTAD)

尽管听力障碍是最常见的感觉疾病，对大约18%的人口产生了巨大的负面影响，但使用听力设备的诊断和康复方法仍然有限，例如，在日常噪声条件下恢复正常的语音感知仍然是不完全的主要是因为关于降低的语音感知的经验知识分散，并且没有系统的数据分析方法，个人从听力设备中受益有限。目前的项目以统计，机器学习指导的方式解决这个问题，结合了基础科学，统计分析和临床听力学，独立于所采用的测试语言：通过建立一个可交换的、易于访问的、可扩展的听力学诊断测量数据库和应用于临床人群的专业的、与语言无关的测试，我们将以精确和个性化的方式分析听力障碍的原因和后果。将使用统计和机器学习方法来确定高度确定性地诊断给定听力障碍所需的最小测量集。这将导致统计上有动机的听觉概况，其将与使用理想化的、基于实验室的“开放式主助听器”验证的个性化治疗建议和听力设备的适配相关联（Grimm等人，2006）与患者自身器械提供的受益相比。因此，仅使用几个特定的听力学测试来验配和评估助听器的益处，将达到一个更全面和有理论基础的听力康复方法，从而提高助听器的效率和接受度。通常，这种统计听力学方法将导致精确和有效的诊断以及现代听力设备的优化处方和拟合。俄罗斯的生物医学数据科学中心（与学术临床研究医院相连）与德国的基础科学驱动的卓越集群之间的特定互动具有建立统计听力学领域的巨大潜力。俄语和德语语言测试的可比性（基于彼得堡和奥尔登堡之间已经存在的合作）以及康复听力学先进程序的联合开发和应用将加强德国和俄罗斯在实际相关的语言交流领域的联系。

Even though hearing impairment is the most common sensory disease with a massive negative impact on approximately 18% of our population, the diagnostics and rehabilitation approaches with hearing devices are still limited, e.g., the restoration of normal speech perception in everyday noisy conditions is still incomplete – primarily because of the scattered empirical knowledge about the reduced speech perception and the limited individual benefit from a hearing device without a systematic data analysis approach. The current project addresses this problem in a statistical, machine-learning-guided way that combines basic science, statistical analysis, and clinical audiology independent from the test language employed: By building up an exchangeable, easily accessible, and expandable database of audiological diagnostic measures and specialized, language-independent tests applied to a clinical population, we will analyze the causes and consequences of hearing impairment in a precise and individualized manner. Statistical and machine-learning methods will be used to identify a minimum set of measures needed for the diagnosis of a given hearing disorder with high certainty. This should lead to statistically motivated auditory profiles which will be linked to individualized treatment recommendations and fitting of hearing devices that are verified using an idealized, laboratory-based “open master hearing aid” (Grimm et al., 2006) in comparison to the benefit provided by the patient´s own device. Hence, using only a few specific audiological tests in fitting and assessing the benefit from hearing devices, a more comprehensive and theory-grounded way of hearing rehabilitation will be reached, thus increasing the efficiency and acceptance of hearing devices. Generally, this statistical audiology approach will lead to a precise and efficient diagnostics and an optimized prescription and fitting of modern hearing devices. The specific interaction between a biomedical data science center (connected to an academic clinical research hospital) in Russia and a basic science-driven Cluster of Excellence in Germany has the great potential to establish the field of statistical audiology. The comparability of speech tests in the Russian and German language (which is based on the already existing collaboration between St. Petersburg and Oldenburg) and the joint development and application of advanced procedures in rehabilitative audiology will strengthen the links across Germany and Russia in the practically relevant field of speech communication.