(PS8-89) Systematic review on the effectiveness of AI-delivered CBT interventions for depressive symptoms

Abstract Text:

Introduction: Cognitive Behavioral Therapy (CBT) is a widely recognized psychotherapeutic approach for managing depressive symptoms, focusing on modifying maladaptive thought patterns through cognitive restructuring and behavioral strategies. With the advancement of artificial intelligence (AI), there is growing interest in using AI to deliver CBT-based interventions as an alternative to those provided by therapists or self-guided programs. These AI-powered systems integrate machine learning, algorithms, natural language processing (NLP), and tailored messaging to personalize interactions and offer real-time therapeutic support. This systematic review explores the effectiveness of AI-delivered CBT interventions in alleviating depressive symptoms.

Methods: A systematic literature search was conducted on Web of Science, PubMed, and PsycINFO on November 6, 2024, adhering to PRISMA guidelines. The study selection was based on the PICOTS framework, including only peer-reviewed randomized controlled trials (RCTs) that evaluated AI-driven CBT interventions for depression. To ensure the reliability of findings, quality assessment measures were applied to assess study rigor.

Results: Findings from 16 studies indicated that AI CBT chatbots demonstrated mixed effectiveness across different populations but were generally ineffective in reducing depressive symptoms among individuals aged 50 and older. Notably, many chatbots that functioned without human involvement still proved beneficial in alleviating depression. Surprisingly, higher degrees of personalization did not enhance chatbot effectiveness, nor did the explicit mention of therapist or psychologist involvement lead to better outcomes compared to other AI-driven chatbots.

Limitations: Several limitations should be acknowledged. Despite the exclusion of pilot studies, many included studies had relatively small sample sizes, potentially reducing statistical power. Additionally, inconsistencies in findings complicated the overall interpretation of results.

Conclusion: This review highlights the need for further research to determine which individuals benefit most from AI-delivered CBT interventions and whether enhancements in personalization, technology, or human involvement could improve outcomes. To gain a clearer understanding of their effectiveness in treating depressive symptoms, larger and more rigorous studies are necessary.