The Co-design of an Embodied-Conversational-Agent-based system to help stroke survivors to manage their recovery: the iTakeCharge study

Stroke is a brain injury and the second leading cause of death globally, with most stroke survivors having physical or psychological sequels. Stroke survivors that actively engage with their own rehabilitation process have better recovery and rehabilitation results. Research has explored the use of technology in support of physical and cognitive rehabilitation, such as gamification or robotics. However, despite the growing use of Embodied Conversational Agents to support patients with healthcare management, their use in self-management rehabilitation for stroke survivors has not been reported.

This MRes thesis presents results of the iTakeCharge study, which uses the conversational agent TaCIA (Taking Charge Intelligent Agent) to facilitate a narrative based self-management intervention for stroke recovery, inspired by the Taking Charge After Stroke (TaCAS) session. In previous studies, TaCAS has been shown to contribute to significant improvements in quality of life, independence, and reductions in disability health. The iTakeCharge study adapted the TaCAS session module ‘Look(ing) at the big picture’ to be deployed in a digital session by TaCIA, as opposed to a human facilitator. The objective of this research is to evaluate the feasibility of a TaCAS inspired digital session facilitated by a conversational agent in support of stroke survivors.

Three pilot studies were conducted with different stakeholders: TaCAS researchers, former TaCAS session facilitators and stroke survivors; each cohort having a session with TaCIA. These participants provided feedback throughout questionnaires, including the System Usability Scale, Working Alliance Inventory and other crafted questions that aimed to evaluate feasibility and acceptability, as well as capture their perceptions and suggested improvements. The iTakeCharge study steps for system design, protocol intervention, feedback collection, system modification, as well as the discussion of the human-computer interaction approach to deliver self-management intervention for stroke recovery are presented in this thesis.