Towards a Natural Language-based Dialog Model as an Assistant Tool for Dementia Therapy

Abstract

Reminiscence therapy is a treatment of dementia that mitigates the unstable psychological states of a person with dementia, through conversations between patient and caregiver that evoke past experiences focusing on the patient's long-term memory. It is recommended that a person with dementia receive this type of therapy constantly. In this work, we propose the design of a dialog model that generates personalized conversations between a conversational system and patients with Alzheimer's disease, with the aim to assist their caregivers in providing this type of therapy. The proposed model will be capable of getting a semantic interpretation about what the user is saying, identifying the type of dialog, also it will use information from the patient's life history and lifestyle to generate the conversations. During the dialog between the system and the patient, the model searches for relevant information related to the patient's history through the use of ontologies to enrich and develop a further conversation.