Virtual Agents for Dementia: Personal Assistant, Trainer and Therapist

Author: Ying Tiong

Tiong, Ying, 2017 Virtual Agents for Dementia: Personal Assistant, Trainer and Therapist, Flinders University, School of Computer Science, Engineering and Mathematics

Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact with the details.


Memory life events and experiences are important information to recall personal details and achievement to document the traits and characteristic that makes someone unique and personal. Memories do decline across adult lifespan , which potentially leads to dementia and this distinctive chronic memory declination in aging group, not only leads a pessimistic view due to disappointment in regaining their memory but also disheartening for the family members as the patients often forgetting the critical moments that cherished and their identities that values. This thesis will focus on a method of significant life events storage by daily life video recording in order to manage their memories in replacement with digital memories of life, helping the memory recovery and data retrieval, and possibly brain training for the dementia for a chance to revive their cognitive function. Challenge in increasing amount of digital “memories” will soon meet the challenge of “information overload”, which requires a better approach to manage and storing the events. Therefore, the research mainly involves studying concept behind reserving significant key frames extraction for digital memory hooks whilst disposing redundant video frames to reserve the digital memory. This thesis covers the literature review studies the human brain that affects the memory in conjunction with consultation from experienced caretakers and medical professionals, collecting data to identify key life events that is crucially important for digital memory hooks and identify technique to distinguish the feature from personal life memory to extract flawless video key frame.

Keywords: memory for life (M4L), digital memory hooks, agents for dementia, video keyframe, assistive technology, cognitive training, memory recovery

Subject: Computer Science thesis

Thesis type: Masters
Completed: 2017
School: School of Computer Science, Engineering and Mathematics
Supervisor: Prof David Powers