Intelligent Devices to Engage Cognitively Disabled Older Adults in Visual Artwork

Speaker: Jesse Hoey, University of Dundee

In this talk, I will present a class of devices, called ePADs, for use by art therapists working with older adults with a progressive illness such as Alzheimer's disease. An ePAD combines a touch-screen interface with intelligent user modeling and sensing through cameras using computer vision. Using a probabilistic model, an ePAD monitors the behaviours of a user as well as aspects of their affective or internal state, including their responsiveness and engagement with the device. The ePAD then uses decision theoretic planning to enable situated, adaptive strategies for interaction with a human user. I will discuss results and analysis of a survey of arts therapists, and of one-on-one interviews, and then give details of the ePAD class, framed as a partially observable Markov decision process, or POMDP. A key element of this class is that instantiations can be easily made for a wide range of customisable devices and interface applications for art-making moderation. I will show examples of particular instances of this model on three devices and with three different interfaces, and will give laboratory demonstrations of the functionality of the devices. I will present and discuss the next steps, including hierarchical modeling of multiple activities, and end user testing.