Premium
TD‐P‐019: Sensing Disorientation of Persons with Dementia in Outdoor Wayfinding Tasks Using Wearable Sensors to Enable Situation‐Aware Navigation Assistance
Author(s) -
Koldrack Philipp,
Teipel Stefan J.,
Kirste Thomas
Publication year - 2016
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2016.06.265
Subject(s) - orientation (vector space) , wearable computer , context (archaeology) , gesture , dementia , computer science , cognition , spatial contextual awareness , activities of daily living , human–computer interaction , reminiscence , cognitive map , motion (physics) , wearable technology , psychology , computer vision , artificial intelligence , cognitive psychology , medicine , disease , paleontology , geometry , mathematics , pathology , neuroscience , psychiatry , biology , embedded system
ease in the US. The Dementia Care Ecosystem defines a proactive model aimed at reducing costs by reducing emergency room use. Methods:Mobility and activity patterns in the home are monitored for sudden changes and undesirable trends using off-the-shelf hardware: Bluetooth home sensors, Android smartwatches, and Android smartphones. Alerts are sent to minimally trained “care team navigators”. Adaptive filtering, supervised learning, and unsupervised learning techniques are used to perform room-level indoor positioning and detect performance of activities of daily living (ADLs). Results:As of February 3, 3016, the mobility functional unit has undergone 9 cumulative months of beta testing and the activity unit is under development. Deployment with patients (n1⁄4100) is expected to begin July 1. Conclusions:A home system for passively monitoring Alzheimer’s disease progression can be designed using off-the-shelf hardware. The technical challenges are indoor positioning, ADL classification, and scaling the technology from prototype to deployment in many homes. Prototype results show 94% accuracy in room locationing, many tradeoffs that must be made concerning battery life, and 3x times expected for scaling. If this technology facilitates cost-effective human-inthe-loop monitoring to reduce ER use will be seen following deployment with patients in July.