Predicting Older Adults' Characteristics Using SARs Through Conversational Interaction: A Mixed-Methods Study

This study examined how older adults engage with socially assistive robots (SARs) using a mixed-methods approach that integrated survey data, behavioral logs, and conversational data. Among 191 participants aged 65 and older, 40.8% actively interacted with the Hyodol SAR. Discriminant analysis identified education level, life satisfaction, and early usage frequency as key predictors of engagement. K-means clustering revealed three distinct user profiles: Social Engagers, who prioritized emotional connection despite low self-esteem; Independent Reflectors, who exhibited self-sufficiency with minimal social interaction; and Emotionally Expressive Users, who formed strong emotional bonds with the Hyodol SAR. In the predictive models, religious affiliation and marital status were key indicators for Social Engagers; smartphone proficiency and Satisfaction With Life Scale (SWLS) for Independent Reflectors; and subjective health status for Emotionally Expressive Users.