SHB: Type I (EXP): Algorithms for Unsupervised and Online Learning of Hierarchy of Features for Tuning Cochlear Implants for the Hearing Impaired

SHB：I 型（EXP）：用于调整听力障碍者人工耳蜗的特征层次结构的无监督和在线学习算法

• 批准号: 1231620
• 负责人: Bonny Banerjee
• 金额: $ 29.82万
• 依托单位: University of Memphis
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2013
• 资助国家: 美国
• 起止时间: 2013-01-01 至 2016-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1231620&HistoricalAwards=false
• 关键词: SHB; Type; EXP; Algorithms; Unsupervised

Since noteworthy events happen only occasionally in any data, it is imperative for smart sensors to learn the norms in data so that authorities can be alerted and appropriate action can be taken at the occurrence of an abnormal or noteworthy event. The aim of this project is to develop algorithms that can learn the norm in terms of a hierarchy of meaningful features from data in an unsupervised and online manner. The application testbed is the problem of automatically tuning cochlear implants (CIs) of patients with severe-to-profound hearing loss by continuously monitoring their speech output. The working hypothesis is that deficiencies in hearing for people with significant hearing loss are reflected in their speech production. This project will develop and use unsupervised, online, and biologically plausible machine learning algorithms to learn feature hierarchies from the speech output data of severely-to-profoundly hearing-impaired patients. The learned feature hierarchy from the speech of a patient will be compared to those learned from the speech of a comparable normal hearing population. Deficiencies in the patient's hearing will be ascertained by identifying the missing or distorted features. Algorithms will be developed to map this information into the signal processing strategies used in CIs to enhance the audibility of speech.The proposed project promises transformative changes to three major interdisciplinary fields: machine learning and artificial intelligence, healthcare, and sensors. It will transform the traditional ways in which the clinical needs of patients are met. For example, the results of this project will provide doctors with evidence-based practices that will better address the specific needs of individual patients by monitoring each patient around the clock at minimal effort and cost.Hearing loss is the most common birth defect in the U.S. with slightly over 15,000 new pediatric cases each year and societal losses amounting to $4.6 billion over a lifetime. A proven technology for CI tuning would make a significant difference to the lives of over 1.2 million CI candidates in the U.S. and many more around the world, thereby leading to substantial health and economic benefits to society. Other than CI tuning, the proposed algorithms will be applicable to a variety of monitoring applications within healthcare, such as blood pressure, cerebrospinal fluid pressure, intracavitary pressure of the bladder, etc., and beyond healthcare, such as web, machine health, traffic, etc. Continuous monitoring with wearable and implantable body sensors will increase early detection of emergency conditions and diseases in at-risk patients and also provide a wide range of healthcare services for people with various degrees of cognitive and physical disabilities. Not only the elderly and chronically ill, but also the families in which both parents have to work will benefit from these systems to provide high-quality care services for their babies and children. Finally, the proposed project will integrate diversity by promoting teaching, learning, and interdisciplinary research among underrepresented groups.

由于值得注意的事件只是偶尔发生在任何数据中，因此智能传感器必须学习数据中的规范，以便当局可以在发生异常或值得注意的事件时发出警报并采取适当的行动。该项目的目的是开发算法，可以以无监督和在线的方式从数据中学习有意义特征层次结构的规范。应用测试平台是通过连续监测重度至极重度听力损失患者的语音输出来自动调谐人工耳蜗（CI）的问题。工作假设是，严重听力损失的人的听力缺陷反映在他们的言语产生中。该项目将开发和使用无监督、在线和生物学上合理的机器学习算法，从轻度到重度听力障碍患者的语音输出数据中学习特征层次。将从患者的语音中学习的特征层次结构与从可比较的正常听力人群的语音中学习的特征层次结构进行比较。通过识别缺失或扭曲的特征，可以确定患者的听力缺陷。算法将被开发来映射这些信息到CI中使用的信号处理策略中，以增强语音的可听度。拟议的项目有望对三个主要的跨学科领域产生变革性的变化：机器学习和人工智能，医疗保健和传感器。它将改变满足患者临床需求的传统方式。例如，该项目的成果将为医生提供基于证据的实践，通过以最小的努力和成本全天候监测每位患者，更好地满足个体患者的特定需求。听力损失是美国最常见的出生缺陷，每年有15，000多例新的儿科病例，一生中的社会损失达46亿美元。一项经过验证的CI调整技术将对美国和世界各地超过120万CI候选人的生活产生重大影响，从而为社会带来巨大的健康和经济效益。除了CI调谐之外，所提出的算法将适用于医疗保健内的各种监测应用，诸如血压、脑脊液压力、膀胱的腔内压力等。使用可穿戴和植入式身体传感器进行持续监测将增加对高危患者的紧急情况和疾病的早期检测，并为具有不同程度认知和身体残疾的人提供广泛的医疗保健服务。这些制度不单止惠及长者和长期病患者，亦惠及父母均须工作的家庭，为他们的婴儿和子女提供高质素的照顾服务。最后，拟议的项目将通过促进教学，学习和代表性不足的群体之间的跨学科研究整合多样性。