个性化老年人碰撞前跌倒探测的研究

Pre-impact fall detection has been reported to be effective in preventing falls in older adults. It can help initiate timely medical assistance for fall victims, and/or activate on-demand fall prevention systems to prevent the serious physical injuries often caused by fall impacts. Appropriate fall detection indicators contribute to the performance of pre-impact fall detection. However, existing literature does not provide any scientific basis for fall detection indicator selection. Besides, the existing fall detection models cannot address inter-individual differences in kinematic patterns as well as the conflict between false alarm and misdetection. This proposal aims to provide scientific basis for fall detection indicator selection, and to explore an individual-specific fall detection solution. In order to achieve these aims, an experimental study is proposed to collect older adults’ whole-body kinematic data during realistic falls and activities of daily living (ADLs). By analyzing the collected kinematic data, kinematic differences between falls and no-fall activities in older adults will be identified. Based on the knowledge of such differences, appropriate fall detection indicators will be proposed. In addition, pre-impact fall detection will be modeled by using computational intelligence approaches and statistical process control charts, respectively. Individual factors will be taken into account during model development. By comparing the fall detection performance of different models, the optimal fall detection model will be suggested.

碰撞前跌倒探测是一种有效防止老年人跌倒的方法，它可以在无人干预下使跌倒者得到及时医疗救助，及激活跌倒预防装置避免跌倒碰撞对身体造成的伤害。选取合适的跌倒探测指数对有效的碰撞前跌倒探测至关重要。但是，现有文献没有为跌倒探测指数选取提供科学依据。同时，现有的碰撞前跌倒探测模型不能解释人体运动规律的个体差异，也不能有效解决‘虚警’与‘误探’之间的矛盾。本项目研究目的是：为跌倒探测指数选取提供科学依据，及探索个性化的跌倒探测模型。为了实现研究目的，本项目：（1）以受控实验为手段，收集老年人在真实跌倒和日常活动中的全身运动数据，通过对全身运动数据进行分析，揭示区别跌倒事件与非跌倒事件的人体运动学机理，以此为基础提出合适的跌倒探测指数；（2）分别从统计过程控制图和计算智能的角度，提出能够反映人体运动规律个体差异的，新颖的碰撞前跌倒探测模型，通过对不同模型探测性能的比较，发现最优的个性化跌倒探测方法。

跌倒是造成意外人体损伤的最主要原因，尤其是在老年人中。本项目的总体研究目标是为了更好的理解跌倒的机理，并提出预防跌倒的解决方案。针对这一目标，开展了一系列的实验研究，解决了若干跌倒预防研发中亟需解决的基础性及应用性研究问题。主要完成的研究包括：研究了年龄的增长对跌倒风险的影响机理；揭示了不同风险因素对跌倒风险的影响机理，通过理解这些机理，确定能够有效反映跌倒风险的生物力学指标；使用惯性测量单元与压力传感器，提出能够有效评估跌倒风险的可穿戴智能鞋垫干预方法；提出并检验了基于机器学习方法支持向量机的跌倒探测模型。我们的主要研究贡献包括：为在不同跌倒风险场景下跌倒探测指数的合理选取提供了实证依据；提出了可以应用于预防跌倒的可穿戴解决方案；提出并检验了新颖的跌倒探测解决方案。这些工作的研究成果帮助很好的完成了本项目的研究目标，为解决跌倒预防问题提供了新的科学依据。

内容获取失败，请点击重试