A cognitive layer architecture to support large-language model performance in psychotherapy interactions - PubMed
7 hours ago
- #Psychotherapy
- #AI in healthcare
- #Cognitive-behavioral therapy
- A cognitive layer architecture enhances large language models (LLMs) for psychotherapy interactions.
- In a randomized evaluation, 227 participants interacted with therapy agents, and 22 expert clinicians assessed the transcripts.
- Augmented LLMs outperformed standalone LLMs and human clinicians in cognitive-behavioral therapy competencies.
- Real-world deployment analysis of 19,674 transcripts showed symptom improvement and higher clinical recovery rates with the cognitive layer.
- The architecture supports high-quality AI-assisted cognitive-behavioral therapy, warranting further research into its mechanisms and efficacy.