Augmenting Motion Sensing to Improve Detection of Periods of Unusual Inactivity

摘要

Two sensing options are examined for their potential to improve the sensitivity of a system that detects periods of inactivity in the homes of elderly persons. A previous prototype used passive infrared motion sensors and door sensors combined with a learning algorithm to detect periods of unusual inactivity such as late wake-ups or the aftermath of a fall. This system worked as intended but suffered from low sensitivity, especially at nighttime, since the motion sensors were not able to distinguish a fall in the bedroom from a person getting into bed. Experiments with a worn accelerometer and with bed and chair occupancy sensors suggest that both can dramatically improve system sensitivity. The optimal solution may depend on the users' activity level, living area size, and willingness to wear a device.