Jialing Wang, Xue Wang, Xiaorong Dong, Xindi Shao, Zhaojun Wang, Xiaojie Xie, Xueyan Zhang
NLP-based chatbots may effectively alleviate older adults' loneliness and depression but have no significant effect on overall quality of life. With respect to loneliness reduction, the intervention effect may be correlated with frequency and duration. Older adults exhibit high acceptability of these chatbots and a strong preference for personalized, anthropomorphic interactions.
BACKGROUND: Loneliness is widespread globally, especially among older adults. Chatbots based on natural language processing (NLP) have demonstrated unique potential for alleviating this problem because of their high accessibility, low cost, and spatiotemporal flexibility; however, systematic reviews that comprehensively assess their effectiveness and feasibility are lacking.
OBJECTIVES: To evaluate the effectiveness and feasibility of NLP-based chatbot interventions systematically in alleviating loneliness among older adults.
METHODS: A systematic search for relevant studies was conducted across 6 databases, namely Web of Science, PubMed, Embase, Cochrane Library, CINAHL, and PsycINFO, from their inception to January 7, 2026. The inclusion criteria were having older participants, natural language chatbot interventions, and loneliness as an outcome. Two researchers independently screened the literature, extracted the data, and assessed the study quality using the Methodological Index for Nonrandomized Studies tool and the Cochrane Risk of Bias tool 2.0.
RESULTS: A total of 11 studies (4 randomized controlled trials and 7 quasi-experimental trials) were included. The results revealed that 6 of 9 studies (6/9; 66.7%) reported significant loneliness reduction, 2 of 4 studies (2/4; 50%) reported improved depression, and no significant overall quality-of-life changes were observed (only partial subdomain improvements). Feasibility was high (attrition ≤20%), with positive satisfaction, although the participants called for better interaction naturalness, personalization, and technical adaptability.
CONCLUSION: NLP-based chatbots may effectively alleviate older adults' loneliness and depression but have no significant effect on overall quality of life. With respect to loneliness reduction, the intervention effect may be correlated with frequency and duration. Older adults exhibit high acceptability of these chatbots and a strong preference for personalized, anthropomorphic interactions.