Artificial Intelligence-Based Socially Assistive Robots in Dementia Care: A Systematic Review and Meta-Analysis

Background:
As dementia rates rise globally, there is an urgent need for innovative, non-pharmacological interventions to support individuals with Alzheimer’s Disease and Related Dementias (ADRD) and their caregivers. Socially assistive robots (SARs), particularly those equipped with artificial intelligence (AI), offer personalized, adaptive interaction and are increasingly recognized as promising tools in dementia care. This study systematically reviewed and meta-analyzed the effectiveness of AI-based socially assistive robots (AI-SARs) across institutional and community-based care settings.

Methods:
We conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) evaluating AI-SAR interventions targeting individuals with ADRD and/or their caregivers. A comprehensive search of 11 databases (e.g., PubMed, PsycINFO, Web of Science) yielded 14 eligible studies for systematic review and 10 for meta-analysis. Included interventions featured AI functionalities such as machine learning, emotional recognition, and real-time adaptation. Outcomes analyzed included cognitive function, depression, anxiety, agitation, and quality of life. Standardized mean differences (SMDs) were calculated and synthesized using random-effects models.

Results:
AI-SAR interventions were associated with significant reductions in depression (SMD = -0.30, 95% CI: -0.56 to -0.03, p = .027) and agitation (SMD = -0.26, 95% CI: -0.49 to -0.03, p = .027). No significant effects were found for cognitive function, anxiety, or quality of life. Most studies were conducted in long-term care settings, with a smaller number evaluating community-based or home-based applications. Despite small sample sizes and methodological heterogeneity, findings suggest that AI-SARs may enhance emotional well-being and reduce behavioral symptoms among individuals with ADRD.

Implications:
This review highlights the therapeutic potential of AI-SARs in alleviating depressive symptoms and agitation in dementia care—two outcomes closely tied to quality of life and emotional well-being. While findings for cognitive and caregiver outcomes remain inconclusive, the observed benefits suggest that AI-SARs can serve as effective adjuncts to traditional care models, particularly in environments with limited staffing or high caregiver strain. These findings carry important implications for gerontological social work by demonstrating how adaptive, emotionally responsive technologies can enhance person-centered care. Moreover, by engaging individuals with ADRD in calming and meaningful interactions, AI-SARs may offer indirect respite for family caregivers—helping to reduce stress and burden. Future research should include large-scale, multisite RCTs with standardized outcome measures and longer follow-up periods. Greater attention to caregiver-specific outcomes, cultural tailoring, health equity, and the accessibility of emerging technologies is essential for informing inclusive social work practice and aging-related policy.