Extended Coupled Probabilistic Timed Automata for Monitoring Eating Activities of Elderly Person  
Author Hanan Nasser Muhajab

 

Co-Author(s)

 

Abstract Elderly people in nursing care, assisted living, hospital-recovery rooms and hospice have many problems with the management of daily activities such as walking, dining and personal hygiene. Many of these tasks cannot be attended regularly due to limited availability of human help. There is a need for automated intelligent systems that can monitor patient’ activities, provide instructions to the patients and communicate with nurses for possible intervention. Such systems will become a reality due to the availability of low-cost cameras and similar devices. In this research, we modeled a new Context Sensitive Coupled Guarded Event Based- eXtended probabilistic time finite automata to build and adapt the motion of the elderly person's dining action such as lifting of cans and glass, spoon motion, putting food in the mouth to ensure that the person is feeding oneself without spilling the food. Actions analyzed based on cooperation between head movement, hand movement, mouth movement, and finger movement along with identifying objects such as glass, spoon and fork. A new variation of Probabilistic Timed Automata (PTA) has been used for the sake of taking care of the actions’ variation due to context change. Motion of each part is modeled, as a finite state in the machine (CGCXPTA) and the state will be either real or phantom.

 

Keywords Probabilistic timed automata, activity recognition, HMM
   
    Article #:  DSIS19-78
 
Proceedings of ISSAT International Conference on Data Science & Intelligent Systems
August 1-3, 2019 - Las Vegas, NV, U.S.A.