Automated Health Assessment through Mobile Sensing and Machine Learning of Daily Activities

项目总结/摘要 世界人口正在老龄化，越来越多的老年人患有阿尔茨海默病， 相关性痴呆（ADRDs）是我们社会必须应对的挑战。虽然医疗保健的未来 服务的质量似乎不确定，同时普适计算和智能 嵌入式系统提供了满足这些需求的创新策略。两个特殊的需求， 可以帮助解决的是早期发现认知和身体衰退，并跟踪新的，健康的整合 大脑行为融入日常生活。Adaptelligence LLC的长期目标是将智能手表商业化 一款名为AcTelligence的应用程序，用于评估一个人的认知和身体健康状况，并促进大脑健康 行为本申请的目的是进行研究和开发，以完善和商业化 智能手表应用程序，提供从智能手表传感器检测日常生活活动的功能， 来自活动标记传感器数据的数字行为标记， 标记，并以健康状态和健康行为提示的形式提供用户反馈。这项技术 是独一无二的，因为我们考虑了一个人的整个行为特征，并引入了机器学习方法， 从这些信息中稳健地预测临床测量。我们利用流行的智能手表平台， 可及性和平衡持续评估与扩大和改善健康的机会。基础上 我们成功的第一阶段的努力，我们的方法是提取活动感知的数字行为标记， 智能手表传感器数据（目标1），基于这些标记自动进行健康评估（目标2），并执行 参与式设计网络仪表板，为大脑健康提供可视化分析和警报（目标3）。我们 将为100名老年人的样本验证传感和机器学习技术，并将完善 通过18名参与者的多轮参与式设计进行互动分析。该应用程序将 通过全面的市场分析和战略性的商业化计划推向市场。 拟议的贡献是重要的，因为它们将提供认知和身体健康的见解 在一个人的日常环境中发现，促进早期发现认知和身体衰退 可以带来更有效的治疗这项工作很重要，因为越来越多的老年人 由于慢性健康状况而经历认知和功能限制的个体。此外，委员会认为， 这项工作解决了个人在自己的工作中尽可能长时间保持功能独立的需要。 家庭，从而提高生活质量和降低医疗保健费用。

PROJECT SUMMARY / ABSTRACT The world's population is aging and the increasing number of older adults with Alzheimer's disease and related dementias (ADRDs) is a challenge our society must address. While the future of healthcare availability and quality of services seems uncertain, at the same time advances in pervasive computing and intelligent embedded systems provide innovative strategies to meet these needs. Two particular needs which technology can help address is early detection of cognitive and physical decline, and tracking integration of new, healthy brain behaviors into everyday life. The long-term goal for Adaptelligence LLC is to commercialize a smartwatch app, called AcTelligence, to assess a person's cognitive and physical health and to promote healthy brain behavior. The objective of this application is to perform research a development to refine and commercialize a smartwatch app that offers capabilities to detect activities of daily living from smartwatch sensors, extract digital behavior markers from activity-labeled sensor data, predict clinical health measures from behavior markers, and provide user feedback in the form of health status and healthy-behavior prompts. This technology is unique because we consider a person's entire behavior profile and introduce machine learning methods to robustly predict clinical measures from this information. We utilize a popular smartwatch platform to increase accessibility and balance continuous assessment with opportunities to extend and improve health. Building on our successful Phase I effort, our approach is to extract activity-aware digital behavior markers from smartwatch sensor data (Aim 1), automate health assessment based on these markers (Aim 2), and perform participatory design of a web dashboard that provides visual analytics and alerts for brain health (Aim 3). We will validate the sensing and machine learning technologies for a sample of 100 older adults and will refine the interactive analytics through multiple rounds of participatory design with 18 participants. The app will be brought to market through a thorough market analysis and a strategically-designed commercialization plan. The proposed contributions are significant because they will provide insights on cognitive and physical health revealed within a person's everyday environment that promote early detection of cognitive and physical decline that can lead to more effective treatment. This work is important because of the increasing number of older individuals experiencing cognitive and functional limitations due to chronic health conditions. Furthermore, the work addresses the need for individuals to remain functionally independent as long as possible in their own homes, thereby improving quality of life and reducing health care costs.