Effectiveness of AI-Based Conversational Agents in Mental Health Interventions: A Meta-Analysis

Abstract

Mental health issues such as stress, depression, anxiety and emotional instability remain to be extensive public health challenges, along with limited access to traditional treatment, especially in low-resource situations. Artificial Intelligence (AI) based chatbots or conversational agents have arisen as cost effective and scalable tools for providing mental health care. This integrative review and meta-analysis synthesize researched from 20 peer-reviewed studies assessing the effectiveness of AI-based chatbots in mental health support. The review encompasses 3 main dimensions: [1] the aggregated effectiveness of chatbots in decreasing symptoms of emotional instability, anxiety, depression, and stress; [2] the role of personalization in improving engagement and therapeutic results; and [3] the relative impact of short-term vs long-term applications. The results indicate that AI chat agents, especially those based on cognitive behavioural therapy (CBT), exhibit reasonable effectiveness in decreasing mental health symptoms, particularly in short-term intervention. Personalization significantly increases both the treatment outcomes and engagement, with higher retention and improvement in symptoms as described by users who received tailor-made interventions. Nevertheless, the literature discloses a significant gap in long-term follow-up data, preventing conclusions about persistent benefits. Although AI conversation agent provides to be effective tools in mental health support, further sequential research is necessary for assessing their long-term clinical usage and integrating into hybrid support models.