P4‐389: CARE SUPPORT SYSTEM USING ONTOLOGICAL MODEL OF CARING PATIENT WITH DEMENTIA

Abstract

Caregivers for Person with Dementia (PwD) have difficulties in choosing the appropriate care method because of the complex behavior and psychological symptoms of dementia. To alleviate such problems, interventions that use Web-based applications have been proposed. For example, some systems are designed to share care knowledge among different caregivers so that caregivers can make suitable care decisions for individual PwD. In this study, we introduce a dementia care support system equipped with an ontological model of caring PwD that could provide various care guidelines through personal computer system. We expanded the ontological model of caring PwD that was previously developed in the lab (Fig 1). In essence, we added more knowledges on activities of daily living, dementia symptoms and related care methods that are required to help PwD during daily living at home. To infer distinct care method for individual PwD with different symptoms, proposed system is equipped with contextual reasoners programmed in Semantic Web Rule Languages (SWRL) and Prolog. Figure 2 shows the overall architecture of the system consisting three different components: perception manager, knowledge system and application module. Contextual data from several sensors is processed by the perception manager. Recognition results and query for care method are transmitted to the knowledge system to retrieve possible care methods, which are provided through the robot platform or knowledge sharing platform in application module. Figure 3 illustrates an example of reasoning process when caregiver's task is to help PwD eat a meal. Through knowledge sharing platform, caregiver can not only provide patient assessment information but also obtain care guide for different task. For example, caregiver can find (1) necessary objects to be prepared for a task, (2) spatial information of objects, (3) temporal events related with current and future task and so on (Fig. 4).