WARE-Care: a novel RF-based system to assess and prevent falling

项目摘要/摘要 在老年人独自一人的夜间，护理设施中的许多跌倒事件都没有报道。它可能会导致 生活质量下降，对跌倒和活动受限的恐惧增加，功能能力下降，以及 增加受伤或死亡的风险。我们建议Ware-Care：基于MmWave的秋季评估和 预防，作为一个非侵入性的系统，在夜间工作，收集和评估老年人的跌倒数据。 Ware-Care将提供准确的移动测量，并且紧凑且易于安装。这个 系统安全、私密、用户友好，适用于护理设施应用。该系统引起了公众的强烈兴趣 护理中心，包括当地的汉德梅克犹太老年服务中心。 Ware-Care系统是基于PI在毫米波雷达上的发展而设计的，新颖而独特 结合先进的机器学习技术。毫米波雷达新领域开启机遇 用于老年人监测，具有与现有视觉和身体佩戴传感器相辅相成的显著优势。 毫米波雷达在所有照明条件下运行，并生成可用于精确 在尊重隐私的同时创建骨架，例如不会收集详细的面部信息。非接触式， 与电池有限且可能会刺激身体的传感器相比，非侵入性传感器在 老年人的睡眠时间。我们将首先在不同的房间构建和验证建议的系统，包括 养老院里典型的洗手间和卧室。然后我们将应用新的基于神经网络的方法来 恢复准确的骨骼运动，检测不同年龄、性别、体重的人的异常行为， 或者身高。我们将首先招募一组健康的成年人(18岁或以上)来建立和培训建议的 系统。该系统将在目标年龄较大的群体(65岁及以上)上进一步验证，并与 可穿戴的运动传感器。已经组建了一个跨学科的团队来开发和执行初步的临床 验证了这一新颖的系统。

Project Summary/Abstract Many falls in nursing facilities are not reported during night when older adults are alone. It can lead to reduced quality of life, increased fear of falling and restriction of activities, decreased ability to function, and increased risk of injuries or death. We propose WARE-Care: mmWave based fall Assessment and pRevEntion, as a non-intrusive system to work during the night to collect and assess older adults’ falling data. WARE-Care will provide an accurate measurement of movement and be compact and easy to set up. The system is safe, private, and user-friendly for nursing facility applications. The system is of strong interest to the nursing centers, including the local Handmaker Jewish Services For Aging center. The WARE-Care system is novel and unique and is based on the PI’s developments in mmWave radar combined with advanced machine learning techniques. The new field of mmWave radar opens opportunities for elders monitoring with significant advantages that complement existing visual and body-worn sensors. mmWave radar operates in all lighting conditions and generates a point cloud that can be used to accurately create a skeleton while respecting privacy, e.g. no detailed facial information will be collected. The non-touch, non-obtrusive sensor can be preferred to battery limited and potentially irritating body worn sensors during sleeping hours for elders. We will first build and verify the proposed system in different rooms including a typical restroom and bedroom in the nursing home. We will then apply new neural-network based methods to recover accurate skeleton movement and detect abnormal behavior of people regardless of age, sex, weight, or height. We will first recruit a group of healthy adults (18 or older) for setting up and training the proposed system. The system will further be validated on a targeted older group (65 and older), and be compared with wearable motion sensors. An interdisciplinary team has been assembled to develop and perform initial clinical validation of this novel system.