Ultra Wideband Fall Detection and Prediction Solution for People Living with Dementia

点击翻译按钮获取中文摘要

Abstract: Older adults with cognitive impairment experience an increased risk of falling than those without cognitive impairment. Unfortunately detecting falls or assessing fall risk among persons living with dementia (PLWD) can be challenging due to difficulties in collecting self-reported information or communicating functional test instructions. The proposed project will develop and test an automated fall detection system using Ultra Wideband (UWB) band technology. The advantage of UWB, along with well-established accelerometer and gyroscope technology, is that it produces a more precise resolution (5-10cm) than Bluetooth (1-5m) or Wi-Fi (5-10m). UWB’s real-time location tracking capacity can enhance fall detection accuracy and context, and with a call alert system reduce response time. In addition, the proposed system will collect rich mobility data to enable the detection of mobility-related fall risk (e.g., changes in gait and balance). Thus, if successful, the proposed system is expected to simplify and enhance mobility-based fall risk, fall detection, and quickly send alerts for PLWD. Building on prior work to develop a fall detection system prototype, this fast-track application proposes two phases moving from lab studies to real-world applications. Phase 1 will test the ability of Theora® 360, a novel fall detection system, to detect simulated falls in a laboratory setting, whether sensor location makes a difference in fall detection accuracy and initial user feedback. Milestones to proceed to the next phase are 90% sensitivity and 90% specificity in fall detection in a laboratory setting and codification of protocols, preliminary algorithms, and data platforms for 3D location processing, motion sensing/ categorization, and fall detection. Phase 2 will assess Theora® 360’s ability to detect mobility-based falls among PLWD and to predict changes in fall risk over time in 60 care-recipient-caregiver dyads living at home. Using a previously established neural network, changes in overall mobility, gait characteristics, and daily routines will be observed to develop algorithms for activity modeling and risk profiling. Feedback on technology use and user satisfaction including recommendations for solution improvement will be obtained through technology records, usability surveys, and interviews at the end of the study. Thus, the study represents a mixed model approach with objective sensor data on a 24/7 basis; functional assessments, survey data assessing sociodemographic, care, and psychosocial factors collected five times throughout the study, and qualitative usability data collected toward the end of the study to better understand the complexity of assessing fall risks and commercialization potential of this new technology among PLWD and their caregivers. Envisioned as a corporate-academic partnership between Clairvoyant Networks and Texas A&M University Center for Community Health and Aging, this proposal draws upon expertise in business, aging, dementia, public health, clinical sciences, computer sciences, and engineering, and will benefit from the input of a distinguished advisory group.