A two-layer hierarchical framework for activity sequence recognition by wearable sensors

As the aging population grows, the elderly care service has become an important part of the service industry in the aging population era. Activity monitoring is one of the most important services in the field of the elderly care service. In this paper, we proposed a wearable solution to provide an activity monitoring service on elders for caregivers. This service monitors restroom activities, such as washing hands, urinating and defecation. In the proposed solution, wireless motion sensors are wore on elder's wrist and waist to measure their body movement. The measured motion data are processed to statistical features and aggregated to cloud servers through gateways. A two-layer hierarchical framework is used for the activity recognition. In the first layer, a preliminary recognition is performed by a supervised Reduced Error Pruning (REP) Tree classifier to detect the transition of the activity. In the second layer, a Variable Order Hidden Markov Model (VOHMM) is proposed to determine the sequence of the activities. The experiment results show that the recognition accuracy is 70 percent. We developed a prototype service App to provide a life log for the recording of the activity sequence. The caregivers can make use of this information to take necessary actions accordingly.