Automated (AI) Analysis of Sleep Disordered Breathing

描述（由申请人提供） 这笔培训补助金旨在促进因杜博士的学术生涯 通过建立在申请人的Ayappa？强大的生物医学工程和 计算机背景和提供全面的多学科培训 这将使她成为一名独立调查员申请人？S 职业目标是进入睡眠生理学的全职学术研究。 拟议的培训活动将包括学术课程， 支持研究计划，并使她接触临床研究的各个方面， 睡眠、神经科学和人工智能技术。她将 由医学博士大卫拉波波特指导，Joyce Walsleben博士和莫里斯·奥哈扬 医学博士，哲学博士、以及计算机科学和神经生理学系。的 计划中的研究的目的是开发一个人工智能系统， 用于识别和量化睡眠呼吸障碍（SDB） 仅基于常规检查期间收集的非侵入性心肺信号 多导睡眠图这将简化、标准化和改善对 SDB和促进这方面的研究。长期目标是 SDB光谱的生理学表征，以便临床上 诊断上呼吸道阻力综合征（UARS），本项目的目的是 至：1.从鼻插管气流信号中提取特征以表征 个体呼吸的阻力/可耐受性状态。这些包括 振幅、吸气流量轮廓、Ti/Ttot和振动的存在， 将被用作神经网络的输入， 与已经通过上呼吸的参考测量值分类的呼吸进行比较， 气道阻力（压力/流量）。2a.将信息从 对个体呼吸进行分类以检测和分类呼吸事件 基于单独的流量信号使用训练的神经网络。2b.评价 包括额外的心肺信号如氧气的效用 饱和度、心率、脉搏传导时间和肋骨/腹部运动 （振幅和相位）在这些事件的检测和分类中的应用。 成功完成培训和研究计划将使博士。 Ayappa独立为睡眠领域的研究做出贡献 physiology.

DESCRIPTION (provided by applicant) This training grant is designed to advance the academic career of Dr. Indu Ayappa by building on the applicant?s strong biomedical engineering and computer background and providing comprehensive multi-disciplinary training which will allow her to become an independent investigator. The applicant?s career goals are to enter full time academic research in sleep physiology. Training activities proposed will include academic course work designed to support the research program and expose her to aspects of clinical research in sleep, neural science and artificial intelligence techniques. She will be mentored by David Rapoport, M.D., Joyce Walsleben, Ph.D., and Maurice Ohayon, M.D., Ph.D., as well as faculty in computer science and neurophysiology. The aims of the planned research are to develop an artificial intelligence system for the identification and quantification of sleep disordered breathing (SDB) based solely on non-invasive cardiopulmonary signals collected during routine polysomnography. This will simplify, standardize and improve the diagnosis of SDB and facilitate research in this area. With the long term goal of physiologic characterization of the spectrum of SDB in order to clinically diagnose upper airway resistance syndrome (UARS), the aims of this project are to: 1. Extract features from the nasal cannula airflow signal to characterize the state of resistance/collapsibility of individual breaths. These include amplitude, inspiratory flow contour, Ti/Ttot and presence of vibration which will be used as inputs to a neural network that will be trained and evaluated against breaths that have been classified by reference measurement of upper airway resistance (pressure/flow). 2A. Incorporate information from this classification of individual breaths to detect and classify respiratory events based on the flow signal alone using a trained neural network. 2B. Evaluate the utility of including additional cardiopulmonary signals like oxygen saturation, heart rate, pulse transit time and rib/abdominal movements (amplitude and phase) in the detection and classification of these events. Successful completion of the training and research program will allow Dr. Ayappa to contribute independently to research in the field of sleep physiology.