Psychological Counseling Cannot Be Achieved Overnight: Automated Psychological Counseling Through Multi-Session Conversations

2025-06-07

AI Summary

Existing LLM-based psychological counseling research is confined to single-session paradigms, failing to capture the temporal dynamics of real-world therapy—namely, multi-turn interaction and progressive symptom alleviation across sessions. To address this, we introduce MusPsy-Dataset, the first multi-session psychological counseling dataset grounded in authentic clinical cases. We further propose MusPsy-Model, the first framework to formalize counseling as a cross-session state evolution process. It integrates a progress-aware dialogue architecture comprising multi-turn state tracking, session-level memory enhancement, progressive goal alignment, and personalized intervention generation. Experimental results demonstrate significant improvements over single-session baselines: +12.6%–19.3% gains in empathy consistency, problem-relief tracking, and intervention appropriateness—marking a substantial departure from the limitations of single-turn modeling.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

In recent years, Large Language Models (LLMs) have made significant progress in automated psychological counseling. However, current research focuses on single-session counseling, which doesn't represent real-world scenarios. In practice, psychological counseling is a process, not a one-time event, requiring sustained, multi-session engagement to progressively address clients' issues. To overcome this limitation, we introduce a dataset for Multi-Session Psychological Counseling Conversation Dataset (MusPsy-Dataset). Our MusPsy-Dataset is constructed using real client profiles from publicly available psychological case reports. It captures the dynamic arc of counseling, encompassing multiple progressive counseling conversations from the same client across different sessions. Leveraging our dataset, we also developed our MusPsy-Model, which aims to track client progress and adapt its counseling direction over time. Experiments show that our model performs better than baseline models across multiple sessions.

Problem

• Addressing multi-session automated psychological counseling needs
• Overcoming single-session limitations in current LLM counseling research
• Tracking client progress across dynamic counseling conversations

Innovation

• Multi-session counseling dataset from real cases
• Dynamic client progress tracking model
• Adaptive counseling direction over time