Stone Erik E, Skubic Marjorie, Back Jessica
Annu Int Conf IEEE Eng Med Biol Soc. 2014;2014:2961-4. doi: 10.1109/EMBC.2014.6944244.
A method for automatically generating alerts to clinicians in response to changes in in-home gait parameters is investigated. Kinect-based gait measurement systems were installed in apartments in a senior living facility. The systems continuously monitored the walking speed, stride time, and stride length of apartment residents. A framework for modeling uncertainty in the residents' gait parameter estimates, which is critical for robust change detection, is developed; along with an algorithm for detecting changes that may be clinically relevant. Three retrospective case studies, of individuals who had their gait monitored for periods ranging from 12 to 29 months, are presented to illustrate use of the alert method. Evidence suggests that clinicians could be alerted to health changes at an early stage, while they are still small and interventions may be most successful. Additional potential uses are also discussed.
研究了一种响应家庭步态参数变化自动向临床医生发出警报的方法。基于Kinect的步态测量系统安装在一个老年生活设施的公寓中。这些系统持续监测公寓居民的步行速度、步幅时间和步幅长度。开发了一个用于对居民步态参数估计中的不确定性进行建模的框架，这对于稳健的变化检测至关重要；同时还开发了一种用于检测可能具有临床相关性的变化的算法。给出了三个回顾性案例研究，对象是那些步态被监测了12至29个月不等的个体，以说明警报方法的使用。有证据表明，临床医生可以在健康变化仍很微小且干预可能最有效的早期阶段收到警报。还讨论了其他潜在用途。
