A Novel System to Detect Falls in Real-life Conditions

 描述（由申请人提供）：福尔斯是老年人（>65岁）致命和非致命伤害的主要原因，每年造成300亿美元的医疗费用。医疗警报装置，通常作为吊坠佩戴，可用于在跌倒时发出求救信号。最近，已经开发了具有自动跌倒检测功能的医疗警报设备。这些设备使用加速度计来检测跌倒，如果佩戴者忘记或无法按下警报按钮，则可以发出帮助信号。这些设备的广泛采用受到未检测到的福尔斯和错误警报的普遍存在的限制，并且受到缺乏医学上可用的研究的限制，这些研究记录了在真实世界环境下商业上可用的跌倒检测设备的灵敏度和错误警报率。BioSensics与跨学科先进运动性能联盟和亚利桑那大学亚利桑那老龄化中心合作，开发了一种医疗警报挂件（ActivePERSTM），通过国家老龄化研究所的第一阶段和第二阶段STTR自动跌倒检测，活动监测和违规警报。ActivePERS是使用来自实验室环境中的模拟福尔斯和模拟日常生活活动的数据开发的。在这种情况下，ActivePERS具有100%的灵敏度和特异性。然而，这些跌倒检测算法尚未在现实世界条件下充分表征。该提案的主要目标是在真实世界的设置中测试ActivePERS，并基于来自真实世界的福尔斯的加速度数据，改进ActivePERS跌倒检测算法，以实现灵敏度和误报率之间的最佳权衡。此外，我们打算将ActivePERS的使用扩展到近福尔斯的检测。近福尔斯的检测可以实现新的结果措施，旨在评估旨在减少老年人福尔斯和近福尔斯的干预措施的有效性。为了实现这些目标，200名居住在社区的老年人将佩戴跌倒检测传感器12个月。传感器将被配置为基于现有算法检测福尔斯，以及记录原始三轴加速度计信号，用于算法改进和开发新算法以检测近福尔斯。将检测到的福尔斯和接近福尔斯与自我报告的福尔斯和接近福尔斯进行比较。鉴于第二阶段SBIR的预算限制，这一雄心勃勃的项目通常是不可能的。然而，目前的提案代表了BioSensics和Partners Healthcare（新英格兰最大的医疗保健提供商）之间的独特合作关系。合作伙伴医疗保健是一个正在进行的，多站点，3000万美元的研究赠款，由患者为中心的结果研究所（PCRI）资助的试验网站，以寻找有效的和基于证据的福尔斯预防策略。通过利用广泛的持续患者招募，并依靠正在进行的研究收集自我报告的跌倒日志，我们将能够在SBIR预算限制范围内实现既定目标。拟议的研究将提供迄今为止最大的真实世界福尔斯数据集，并使BioSensics能够将可靠的跌倒检测技术商业化。

 DESCRIPTION (provided by applicant): Falls are the leading cause of fatal and nonfatal injuries among older adults (>65 years) and result in $30 billion in annual medical costs. Medical alert devices, commonly worn as a pendant, can be used to signal for help in the event of a fall. More recently, medical alert devices with automatic fall detection functionality have been developed. These devices use accelerometry to detect a fall and can signal for help if the wearer forgets to, or is incapable of, pressing the alert button. Widespread adoption of these devices has been limited by the prevalence of undetected falls and false alerts, and by the lack of publically available studies documenting the sensitivity and false alarm rate of commercially-available fall detection devices under real-world settings. BioSensics, in collaboration with the Interdisciplinary Consortium on Advanced Motion Performance and the Arizona Center on Aging at the University of Arizona, developed a medical alert pendant (ActivePERSTM) with automatic fall detection, activity monitoring, and non-compliance alerts through a Phase I & II STTR from the National Institute on Aging. ActivePERS was developed using data from simulated falls and simulated activities of daily living in a laboratory setting. In this setting, ActivePERS has 100% sensitivity and specificity. However, these fall detection algorithms have not been adequately characterized under real-world conditions. The primary objectives of this proposal are to test ActivePERS in a real-world setting, and to improve the ActivePERS fall detection algorithm to achieve an optimal trade-off between sensitivity and false alarm rate, based on acceleration data from real-world falls. In addition, we intend to extend the use of ActivePERS to the detection of near falls. The detection of near falls could enable novel outcome measures aimed to evaluate the effectiveness of interventions designed to achieve a decrease in falls and near falls in older adults. To accomplish these objectives, 200 community-dwelling older adults will wear a fall detection sensor for a period of 12 months. The sensor will be configured to detect falls based on existing algorithms, as well as to record raw tri-axial accelerometer signals for th purposes of algorithm improvements and development of a novel algorithm to detect near falls. Detected falls and near falls will be compared to self-reported falls and near falls. This ambitiou project would not typically be possible given the budget constraints of a Phase II SBIR. However, the present proposal represents a unique partnership between BioSensics and Partners Healthcare (the largest healthcare provider in New England). Partners Healthcare is a trial site in an ongoing, multi-site, $30 million research grant, funded by the Patient-Centered Outcomes Research Institute (PCORI), to find effective and evidence-based strategies for falls prevention. By leveraging the extensive ongoing patient recruitment and relying on the ongoing study for collection of self-reported fall logs, we will be able to achieve the stated objectives within the SBIR budget constraints. The proposed study will provide the largest dataset to date of real-world falls and uniquely position BioSensics to commercialize a reliable fall detection technology.