1. Novel Approach to Unsupervised Mobility Assessment Tests: Field Trial For aTUG
- Author
-
Myriam Lipprandt, Andreas Hein, Thomas Frenken, Melina Brell, Elisabeth Steinhagen-Thiessen, S. Wegel, and Mehmet Gövercin
- Subjects
Computer science ,business.industry ,Reference data (financial markets) ,Pattern recognition ,Sensor fusion ,Test (assessment) ,law.invention ,Home automation ,law ,Component (UML) ,Range (statistics) ,Unsupervised learning ,Computer vision ,Artificial intelligence ,business ,Stopwatch - Abstract
A novel approach to performing unsupervised mobility assessment tests in domestic environments is presented. As a part of the aTUG concept the approach is based on the idea to segment assessment tests into components made up of recurring movement patterns which are measured independently by use of ambient sensor technologies. Quality criteria are defined which compute a score of eligibility for usage of sensor data to assess a certain test component. Valid component measurements are recombined to complete assessment tests according to a technical assessment test description defining the flow of segments and their constraints. An experiment has been conducted within a field trial with five elderly people aged 64-84 years over five weeks. The flats of all people were equipped with home automation (HA) sensors. A laser range scanner (LRS) was placed in one flat. Results from the fully-equipped flat show that the presented quality criteria are suitable to select LRS measurements according to their eligibility to assess a certain component. HA sensors and the LRS were used to compute a self-selected gait velocity of 0.71m/s unsupervised at home. TUG using the aTUG apparatus and a stopwatch was used as clinical reference data yielding a mean gait velocity of 1.18m/s. For the described setting a difference of 0.47m/s between capacity and performance in gait velocity was found.
- Published
- 2012