Effectiveness of AI Care Robots In Reducing Social Isolation, Loneliness, and Depression Among Older Adults Living Alone: An Explanatory Sequential Mixed Methods Study

Background and Purpose

As societies transition into super-aged populations, the rapid growth of older adults living alone has intensified a critical “triad of emotional vulnerability”: social isolation, loneliness, and depression. These interrelated conditions reflect not only individual distress but also structural gaps in community-based care systems. AI care robots have emerged as a promising, scalable form of technology-mediated support; however, rigorous evidence—particularly using longitudinal and mixed methods approaches—remains limited. This study evaluates the effectiveness of an AI care robot (“Hyodol”) as a psychosocial intervention and examines the mechanisms through which emotional change occurs. The study is theoretically grounded in the Convoy Model of Social Relations and Sullivan’s Interpersonal Theory, advancing the proposition that AI-human interaction can function as a supplementary support system and activate core interpersonal processes underlying emotional well-being.

Methods

An explanatory sequential mixed methods design was employed. In the quantitative phase, a quasi-experimental pretest–posttest control group design was conducted with 60 older adults living alone (intervention n = 30; control n = 30). The intervention group engaged in daily interaction with the AI care robot over an eight-week period, while the control group received usual community services. Outcomes included social isolation, loneliness, and depressive symptoms, measured using validated instruments (LSNS-6, UCLA Loneliness Scale, GDS). Linear mixed models (LMM) were used to estimate group-by-time interaction effects while accounting for within-subject correlations and covariates. In the qualitative phase, in-depth semi-structured interviews were conducted with a purposive subsample of intervention participants (n = 3). Thematic analysis was used to identify mechanisms explaining the quantitative findings.

Results

Quantitative analyses revealed significant group-by-time interaction effects across all outcomes, indicating greater improvements in the intervention group compared to controls: social isolation (β = −2.54, p = .02), loneliness (β = −0.85, p = .04), and depression (β = −3.00, p = .03). Qualitative findings identified three interrelated mechanisms underlying these effects: (1) restoration of social presence, whereby routine voice interactions reduced perceived isolation by disrupting prolonged silence; (2) perception of emotional companionship, as participants developed affective bonds with the robot, often describing it in relational terms (e.g., as a family-like presence); and (3) recovery of daily routines and self-efficacy, as structured prompts supported behavioral activation and a renewed sense of purpose. The convergence of findings indicates that observed improvements reflect substantive psychosocial change rather than measurement artifacts.

Conclusions and Implications

This study provides mixed-methods evidence that AI care robots can effectively reduce multiple, co-occurring emotional vulnerabilities among older adults living alone. Conceptually, the findings extend the Convoy Model by introducing AI as a form of “digital convoy” and demonstrate that interpersonal regulatory mechanisms can be activated through human–AI interaction. Practically, the results support integrating AI care robots into community-based aging-in-place strategies as a scalable supplement to human services. Policy implications include prioritizing access for high-risk subgroups, particularly socially isolated older women, and expanding public investment in technology-mediated care to address emerging gaps in rapidly aging societies.