Evaluating Older Adults’ Engagement and Usability with AI Driven Interventions (Preprint)

BACKGROUND Technologies that serve as assistants are growing more popular to entertain and help with daily tasks. The artificial intelligence (AI) in these technologies could also be helpful to deliver interventions that assist older adults with symptoms or self-management. To create the best delivery and dose of an intervention, we must understand older adults’ usability and engagement behaviors. OBJECTIVE The purpose of this analysis was to describe how older adults engaged with a commercially available AI assistant that delivered prescribed routines as interventions. METHODS A randomized pilot trial was conducted for 12-weeks in adults aged 60 years or older, self-reported living alone, and having chronic musculoskeletal pain. Participants (N=50) were randomly assigned to one of two intervention groups to engage with the AI assistant Alexa. One group interacted with a standard intervention with a priori activities, and the other group interacted with enhanced (tailored) intervention based on participant input. Participants were encouraged to interact with Alexa routines prescribed twice daily (morning and evening) and as needed. Data were collected and analyzed on routine engagement characteristics and perceived usability of the AI assistant. To understand how to improve the personalization of the user experience, an analysis was conducted on participant’s personality trait scores. RESULTS Participants were on average 79 years of age. In both intervention groups, morning routines were initiated more frequently than evening routines. The enhanced group perceived routine usability as good, and the standard group reported lower usability scores. Higher extraversion personality scores predicted higher rates of routine initiation, while higher agreeableness and higher conscientiousness personality scores positively predicted better usability. CONCLUSIONS These findings support the feasibility of using prescribed interactive AI assistant routines with older adults. Routine engagement and usability may be enhanced with the development of skills and interactive features for aging populations. Older adults' engagement with prescribed interventions with AI assistants may be influenced by personality traits, preferences for routines, and symptom characteristics. CLINICALTRIAL NCT05387447