A social robot to support people with dementia at mealtimes: Design, implementation and preliminary evaluation. (20th December 2022)

• Record Type: Journal Article
• Authors: Astorga, Misael; Favela, Jesus; Cruz‐Sandoval, Dagoberto; Riek, Laurel D
• Is Part Of: Alzheimer's & dementia. Volume 18(2022)Supplement 2
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-12-20
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.061311
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: People with dementia often experience difficulties eating at mealtimes, which is a common cause of concern to informal caregivers. These difficulties include becoming agitated or angry at mealtimes, taking food from someone else's plate, or refusing to eat the food being offered. These behaviors manifest differently in each individual, are context‐dependent, and vary overtime. Method: We engaged in a user‐centered design process to create a social robot that can serve as an eating companion for people with dementia (PwD). Interviews with caregivers (n=6) helped identify common problematic behaviors during mealtimes, and strategies to address them. Additionally, we conducted iterative, interaction co‐design sessions to define the robot's behaviors. The EVA robotic platform was used to implement the PwD‐robot interaction and to create videos of short interactions to show how the social robot can help to deal with these problematic behaviors. These videos were shown to informal caregivers (n=14) who were asked to assess user adoption using the Almere model (Heerink et al. 2010). A machine learning model was trained to detect disruptive eating behaviors from images captured by the social robot and to trigger appropriate responses. Result: Caregivers found the scenarios to be realistic and perceived the robot to be enjoyable and easy to use. A 92% precision was obtained from the trained Machine Learning model in detecting that an individual is eating (83% sensitivity and 96% specificity), using a capture rate of one frame per second. Conclusion: The results provide evidence of the technical feasibility of implementing a social robot to act as an eating companion for people with dementia and the positive perception of caregivers regarding its potential adoption.