Systems for control and evaluation of advanced assistive technologies

康复技术旨在恢复残疾人的独立性和生活质量。颈脊髓损伤后，手臂和手的功能恢复是重中之重。为了满足这一需求，可以恢复各种复杂运动的技术正在取得进展，包括植入功能性电刺激系统和可穿戴外骨骼。然而，人机界面需要相当大的改进，使用户能够有效地控制这些新的辅助技术可以产生的复杂运动模式。与神经系统的接口是最吸引人的途径，因为它们提供了一种直观和自然的控制机制。拟议的研究计划的目标是推进人机界面技术，使复杂的辅助技术的有效沟通。我们将解决以下目标：目标1 -接口：开发方法来选择性地记录来自外周神经内的多个来源的信息，以便从神经系统中提取命令信号。基于我们最近在急性动物实验中的成功，我们将使用慢性动物研究和计算建模来开发自适应信号处理方法，以补偿随时间的变化。这项工作将产生稳定和准确的神经接口，支持下一代辅助技术。目标2 -控制：优化通过部分受损的神经通路可实现的信息传输速率（ITR）。使用来自SCI患者上肢肌肉的肌电图记录作为实验模型，我们将开发方法来预测个性化的控制参数（类的数量，命令的速率），基于所使用的肌肉的剩余强度来优化每个人的ITR。目标3 -评价：开发可穿戴系统，以表征新技术对SCI患者日常生活中上肢功能的影响。我们的团队率先使用可穿戴摄像机（自我中心视频）来评估家中的手部功能。在授予期间，我们将开发计算机视觉和深度学习方法，用于以自我为中心的视频序列中手部的姿势跟踪，从而提供前所未有的手部功能细节。这项工作将催化新一代康复辅助技术。对自然科学和工程的具体贡献将包括新颖的慢性神经接口，这些接口的最佳控制方案，以及直接评估辅助技术的真实影响的可穿戴系统。我们将培养神经工程、信号处理、计算机视觉、深度学习、建模和神经科学方面的高素质人才。这项研究将提高残疾人的生活质量，并减少对医疗保健系统的经济影响。

Rehabilitation technology aims to restore independence and quality of life to individuals with disabilities. After cervical spinal cord injury (SCI), arm and hand function are the top priority for recovery. To address this need, progress is being made in technologies that can restore a variety of complex movements, including implanted functional electrical stimulation systems and wearable exoskeletons. However, considerable improvements are needed in human-machine interfaces that will enable users to effectively control the complex movement patterns that these new assistive technologies can produce. Interfaces with the nervous system are the most appealing avenue, as they provide a mechanism for intuitive and naturalistic control. The goal of the proposed research program is to advance human-machine interface technology to enable effective communication with complex assistive technologies. We will address the following objectives: Objective 1 - Interface: Develop methods to selectively record information from multiple sources within peripheral nerves, in order to extract command signals from the nervous system. Building on recent our success in acute animal experiments, we will use chronic animal studies and computational modeling to develop adaptive signal processing methods that can compensate for changes over time. This work will result in stable and accurate neural interfaces that will support the next generation of assistive technologies. Objective 2 - Control: Optimize the information transfer rates (ITRs) achievable through partially impaired neural pathways. Using electromyographic recordings from upper limb muscles of individuals with SCI as an experimental model, we will develop methods to predict personalized control parameters (number of classes, rate of commands) that optimize ITR for each person, based on the residual strength of the muscles used. Objective 3 - Evaluate: Develop wearable systems to characterize the impact of new technology on upper limb function in the daily lives of individuals with SCI. Our group has pioneered the use of wearable cameras (egocentric video) to assess hand function at home. Over the granting period, we will develop computer vision and deep learning methods for postural tracking of the hands in sequences of egocentric video, thus providing an unprecedented level of detail about hand function. This work will catalyze a new generation of assistive technology in rehabilitation. Specific contributions to the natural sciences and engineering will include novel chronic neural interfaces, optimal control schemes for these interfaces, and wearable systems that directly evaluate the true impact of assistive technology. We will train highly qualified personnel in neural engineering, signal processing, computer vision, deep learning, modeling, and neuroscience. This research will lead to improved quality of life for individuals with disabilities, and reduce the economic impact on the health care system.