Fostering Emotional Engagement and Sustained Interaction in Care Homes Through Empathetic Human-Robot Conversation

Social isolation is a critical health concern for older adults — especially in institutional care, where limited social interaction is linked with depression, cognitive decline, and poorer physical health. Empathetic human-robot interaction offers a promising avenue to address these emotional and mental health challenges. We report on a two-stage evaluation of an LLM-powered social robot designed to support emotionally attuned conversations with older adults. In a lab study ( \(N=40\) ), the robot achieved high scores on four empathy-related measures. We then deployed a fully autonomous system for 10 weeks in a residential care home (five residents; up to 45 minutes per session). Across repeated interactions, participants showed sustained engagement and predominantly mid-depth self-disclosure, alongside affective trust and occasional cognitive trust. Exploratory analyses indicated that higher empathic appropriateness co-occurred with deeper disclosure and increased over sessions for most participants. To our knowledge, these findings provide early field evidence that autonomous, LLM-driven conversational robots can maintain emotionally supportive dialogue with older adults, fostering emotional engagement, trust, and sustained interaction over time, mechanisms that may contribute to reducing perceived social isolation.