| Sumario: | Background and Objectives Family caregivers face elevated risks for mental and physical health issues but have difficulty accessing informational and supportive resources. Chat-based artificial intelligence (AI) tools, such as chatbots using large language models (LLMs), may offer scalable, personalized support. This scoping review maps the existing literature on chatbots for family caregivers. Research Design and Methods Following our preregistered protocol and following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) checklist, we searched PubMed, Embase, Web of Science, and CINAHL (January–February 2025) for English-language, peer-reviewed articles on AI/chatbots and caregivers. Eligibility included any article on the use of a chatbot in the caregiving context. Screening and data extraction were performed in duplicate using Covidence. Articles were charted by article type ((1) existing chatbots [e.g. ChatGPT] applied to caregiving context; (2) new chatbots developed for caregiving context; (3) "other" article types) and synthesized narratively. Results Of 505 records screened, 30 met inclusion criteria. Seventeen evaluated existing chatbots (primarily ChatGPT), nine described development of new chatbots, and four were commentaries, reviews, or qualitative studies. Most focused on dementia or cancer caregiving. The majority of new or existing chatbot articles focused on provision of informational support, and less often on emotional support or local resources. Chatbot evaluations were typically conducted by experts assessing response accuracy and readability; few reported evaluations by caregivers. Discussion and Implications Evidence suggests that chatbots provide largely accurate, accessible caregiving-related information, though findings for emotional support and safety were limited and mixed. Research remains early-stage, with no randomized controlled trials.
|
|---|