Artificial Intelligence-based Pets in Mental Health and Chronic Disease Care: A Critical Narrative Review of Evidence, Mechanisms and Implementation

Abstract

Background: Artificial intelligence-based pets have been developed as alternatives to live-animal companionship in settings where allergies, infection control, animal welfare and care demands restrict access to animals. This critical narrative review evaluates evidence on their mental-health and chronic-disease applications, proposed mechanisms and implementation requirements. Methods: Peer-reviewed studies and authoritative institutional sources were identified through structured searches of biomedical and multidisciplinary scholarly indexes and citation tracking. Randomised and non-randomised trials, systematic reviews, qualitative studies, experimental work and ethical analyses were appraised thematically. Results: Evidence is most developed in dementia care, where robotic pets can improve short-term engagement and positive affect and may reduce agitation in some residents. Findings for depressive symptoms, cognition, quality of life and sleep remain inconsistent. Community studies suggest potential reductions in loneliness, although many are uncontrolled or rely on self-report or proxy assessment. Evidence in chronic pain, haemodialysis, cancer treatment, haematopoietic stem cell transplantation and persistent post-viral illness remains preliminary and is based mainly on small studies. Experimental findings indicate that interactivity contributes to engagement and some sensory outcomes, while simpler non-interactive comparators can produce overlapping affective effects. Conclusion: Artificial intelligence-based pets should currently be considered facilitator-supported adjuncts rather than validated standalone treatments. Future research requires adequately powered, longer-term, mechanism-informed studies that address individual preference, cost, equity, privacy, governance and the durability of effects.