The role of LLM-powered chatbots in assisting elderly people: systematic review

Tahir H., ÖZDEMİR M. K., ALHAJJ R.

Network Modeling Analysis in Health Informatics and Bioinformatics, cilt.15, sa.1, 2026 (ESCI, Scopus)

Özet

The growing elderly population presents significant challenges in healthcare, mental well-being, and social engagement. While traditional interventions have been effective to some extent, there is an increasing need for scalable, personalized, and intelligent support systems. Large Language Model (LLM)-powered chatbots offer a promising solution by providing real-time, adaptive, and context-aware assistance. This systematic review investigates the applications, benefits, and limitations of LLM-powered chatbots in supporting elderly individuals across different domains. Following the PRISMA framework, we conducted a systematic review of studies from 6 academic databases. The selection process included peer-reviewed journal articles, conference papers, and empirical studies that explored the deployment of LLM-based chatbots in eldercare. Inclusion criteria focused on studies demonstrating chatbot applications in patient education, mental health, physical health support, and AI perception. We extracted data on chatbot functionalities, effectiveness, and challenges while assessing study quality through standardized methodological criteria. A total of 46 studies met the inclusion criteria, highlighting diverse implementations of LLM-powered chatbots in elderly care. Chatbots demonstrated efficacy in patient education by providing accessible and evidence-based medical information, improving adherence to treatment regimens, and assisting with chronic disease management. In mental health support, chatbots were found to reduce loneliness, facilitate reminiscence therapy, and support guided self-reflection. For physical health, AI-driven assistants enhanced rehabilitation programs, exercise adherence, and post-surgical recovery. Public perception studies revealed mixed attitudes toward AI-based chatbots, with acceptance influenced by factors such as digital literacy, trust, and privacy concerns. LLM-powered chatbots represent a transformative technology with significant potential to enhance eldercare. Their ability to provide continuous, scalable, and interactive support positions them as a valuable complement to traditional healthcare and social interventions. However, challenges related to accuracy, ethical concerns, personalization, and integration with existing healthcare frameworks remain. Future research should focus on improving chatbot emotional intelligence, ensuring fair and unbiased AI responses, and fostering user trust through transparency and regulation.