Development and Feasibility Studies of AI-powered Socially Assistive Robotics to Promote Wellbeing of Persons with Alzheimer’s Disease and Related Dementias

The number of persons living with Alzheimer's Disease and Related Dementias (PLWDs) has been keeping growing. In 2024, it is estimated that there will be approximately 6.7 million individuals living with Alzheimer's Dementia. This number will increase to about 14 million in 2060. Due to the damage in neurons, the capabilities of memory, thinking, and language will decline as the disease progress. As a result, persons with dementia will gradually withdraw from their social activities and become more dependent on others during their activities of daily living. Making it worse, our society is not ready for the increasing requirements of dementia care workforce, for both informal caregivers and professional caregivers. To support and augment dementia care, my dissertation has been focusing on designing and developing socially assistive robotics to care for people with dementia, though companionship, entertainment, assistance, rehabilitation, and monitoring.

A robot-mediated reminiscence therapy have been investigated. We firstly evaluated the acceptability, feasibility, and user experience of this robot functions with people with dementia and their caregivers. Moreover, we developed reinforcement learning algorithm for the robot to provide adaptive and personalized reminiscence therapy, tailoring to PLWD's affective states and cognitive capabilities.

At the same time, my dissertation has developed the robotic system to guide and assist PLWD during activities of daily living. Since there is lack of accessible dataset, we used tele-operation to control the robot to guide PLWD performing the tasks. We conducted a pilot study of this tele-operated robotic system with people with mild cognitive impairment and dementia. Through this preliminary study, we have modified it and developed a portable robotic platform, which can be easily transferred to other locations and be used to explore PLWD-robot interaction. Furthermore, with collaborations with occupational therapists, we developed a holistic framework, as a benchmark, to evaluate PLWD-robot interaction. This interdisciplinary approach underscores the potential of robotic technologies to meet the intricate needs of PLWDs, advocating for more inclusive and adaptable dementia care solutions. Furthermore, my work extends into employing reinforcement learning to enable SAR to offer adaptive assistance for daily activities, further emphasizing the innovative and user-centered design of our approach.