Mixed methods, single case design, feasibility trial of a motivational conversational agent for rehabilitation for adults with traumatic brain injury

Objective: Rehabilitation for adults with traumatic brain injury (TBI) incorporates client-centred goal-setting and motivational support to achieve goals. However, face-to-face rehabilitation is time-limited. New therapy approaches which leverage care are warranted. Conversational agents (CAs) offer a human-computer interface with which a person can converse. This study tested the feasibility, usability and acceptability of using a novel CA - RehabChat - alongside brain injury rehabilitation. Design: Mixed methods, single case design, feasibility pilot trial. Setting: Ambulatory and community brain injury rehabilitation. Participants: Adults with TBI receiving brain injury rehabilitation and clinicians providing this care. Intervention: Following 1:1 training, client-clinician dyads used RehabChat for two weeks alongside usual care. Main measures: Pre-post clinical measures (Motivation for Traumatic Brain Injury Rehabilitation Questionnaire, Rehabilitation Therapy Engagement Scale, Brain Injury Rehabilitation Trust Motivation Questionnaire-Relative, Brain Injury Rehabilitation Trust Motivation Questionnaire-Self) repeated measures (Hospital Anxiety and Depression Scale, researcher-developed wellbeing screening questions); and post-intervention (System Usability Scale (SUS), semi-structured 1:1 interview). Results: Six participants (two clients and four clinicians) completed training. Two client-clinician dyads completed the intervention. Two other clinicians used RehabChat in a mock client-clinician session. SUS scores indicated good usability. Client well-being did not deteriorate. No adverse events were experienced. Interviews indicated RehabChat was feasible, acceptable and easy to use; and supported motivation, goal-setting and completing practice activities. Conclusions: RehabChat was feasible and acceptable to use alongside usual ambulatory and community brain injury rehabilitation, had good usability and supported client needs. Further testing of RehabChat with a larger cohort for longer duration is warranted.