Emergent event detection and modeling for bedside activity recognition

Due to the rapid increase in older population, there is a growing need to develop a corresponding technology for caregiving. We have implemented a prototype system to detect several abnormal events for bedside activities. This paper proposes the event detection and modeling mechanism used in the system. The mechanism is based on a hierarchical concept: a state is composed of some sensor data and the composition or transition of states causes an event. The design of the prototype and some applied scenarios are also presented in this paper. When abnormal events can be accurately detected prior to the occurrence of accidents, elders would live safer while families and care givers could feel relieved.