Automated objective outcome measures for clinical use in dysarthria

摘要/概要 无法进行口头交流是所有人类条件中最令人衰弱的。在 交流障碍领域，言语语言病理学家（SLP）对质量的感知评估 是评估和记录治疗进展的黄金标准。然而，在这方面， 几十年的研究已经证实，言语的认知-知觉判断是固有的偏差， 这损害了可靠性。原因是人类的感知系统是自适应的， 在多个治疗阶段与个体合作或与患者合作产生偏倚 一个职业生涯中的人群。因此，为了客观可靠地记录治疗结果，有必要 涉及多个不熟悉的听众这在大多数临床环境中是站不住脚的，这意味着主观的 由治疗的临床医生做出印象。对主观评价的依赖直接破坏了 临床实践的质量和临床医生证明干预有效性的能力。 Aural Analytics开发了新的客观声学语音指标，可以可靠地测量语音， 患有神经系统疾病的人群。它的技术是基于一个强大的科学前提， 早期被制药公司和神经学家用于临床研究。听觉分析已收集 并使用其技术分析了数万个语音样本，结果表明， 它的措施是强大的，可靠的，更敏感的纵向变化的讲话比其他 现有的成果措施。我们成功地完成了第一阶段SBIR项目，目的是将我们的 技术应用于SLP临床实践。这个第二阶段的建议自然是建立在我们以前的工作， 在I期完成的基于应用程序的自动化结局测量， 基准。具体而言，SA1将验证针对美国人的听觉分析语音措施， 言语-语言-听力协会（ASHA）的运动功能沟通测量（FCM） 语音;句子可懂度测试;以及专家对语音特征的评级。此外，年龄和 将从600名新的健康人的数据收集中获得所有客观措施的基于性别的规范。 参与者在SA2中，Aural Analytics将与执业语音语言病理学家进行可用性研究 以评估真实的世界效用并改进用户体验。本提案的交付成果将是 全功能的移动的应用程序，通过在临床环境中练习SLP进行验证，具有真实的时间语音 根据现有社区接受的措施验证成果指标。这将导致客观 符合专业标准的工作流程，从而加快我们的道路， 商业化

Abstract / Summary The inability to engage in spoken communication is among the most debilitating of all human conditions. In the field of communication disorders, a speech-language pathologist’s (SLP’s) perceptual evaluation of the quality of speech production is the gold standard for assessment and for documenting treatment progress. However, decades of research have confirmed that auditory-perceptual judgments of speech are inherently biased, which compromises reliability. The reason is that the human perceptual system is adaptive, with perceptual bias accrued by working with an individual across multiple treatment sessions, or by working with patient populations across a career. Thus, to reliably document treatment outcomes subjectively, it is necessary to involve multiple, unfamiliar listeners. This is untenable in most clinical settings, which means that subjective impressions are made by the treating clinicians. The reliance on subjective evaluation directly undermines the quality of clinical practice and a clinician’s ability to demonstrate the efficacy of an intervention. Aural Analytics has developed new objective acoustic speech metrics that reliably measure speech in populations with neurological disorders. Its technology is based on a strong scientific premise and has been adopted early by pharmaceutical companies and neurologists in clinical research. Aural Analytics has collected and analyzed tens of thousands of speech samples using its technology, and the results are demonstrating that its measures are robust, reliable, and more sensitive to longitudinal changes in speech than are other existing outcome measures. We successfully completed a Phase I SBIR project with the aim of translating our technology to SLP clinical practice. This Phase II proposal naturally builds on our previous work by connecting the automated app-based outcome measures completed in Phase I to three complementary clinical benchmarks. Specifically, SA1 will validate the Aural Analytics speech measures against the American Speech-Language-Hearing Association’s (ASHA) Functional Communication Measures (FCM) for motor speech; the Sentence Intelligibility Test; and expert ratings of speech characteristics. In addition, age and gender-based norms for all objective measures will be obtained from collection of data from 600 new healthy participants. In SA2, Aural Analytics will conduct a usability study with practicing speech-language pathologists to assess real world utility and refine the user experience. The deliverable of this proposal will be a fully-functional mobile application, validated by practicing SLPs in a clinical setting, with real time speech outcome metrics validated with respect to existing community-accepted measures. This will result in objective outcomes that fit into the workflow of the professional standard, thereby expediting our path to commercialization.