From Speech to Profile: A Protocol-Driven LLM Agent for Psychological Profile Generation

2026-04-11

AI Summary

This study addresses the challenges of context forgetting and large language model hallucination in the automatic generation of structured psychological profiles from counseling audio. To mitigate these issues, the authors propose StreamProfile, a streaming framework grounded in the PM+ psychological intervention protocol. StreamProfile processes speech transcripts incrementally, constructs a hierarchical evidence memory, and incorporates a protocol-driven, traceable chain-of-thought reasoning mechanism to ensure that every statement in the generated profile is rigorously anchored to verifiable evidence from the original audio. Experimental evaluation on real-world adolescent counseling data demonstrates that the proposed approach significantly suppresses hallucinations and enhances both the factual accuracy and clinical credibility of the generated psychological profiles.

Technology Category

• Natural Language Processing: Generation
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Data Mining & Knowledge Management: Data Stream Mining

Application Category

• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

The psychological profile that structurally documents the case of a depression patient is essential for psychotherapy. Large language models can be applied to summarize the profiles from counseling speech, however, it may suffer from long-context forgetting and produce unverifiable hallucinations, due to overlong length of speech, multi-party interactions and unstructured chatting. Hereby, we propose a StreamProfile, a streaming framework that processes counseling speech incrementally, extracts evidences grounded from ASR transcriptions by storing it in a Hierarchical Evidence Memory, and then performs a Chain-of-Thought pipeline according to PM+ psychological intervention for clinical reasoning. The final profile is synthesized strictly from those evidences, making every claim traceable. Experiments on real-world teenager counseling speech have shown that the proposed StreamProfile system can accurately generate the profiles and prevent hallucination.

Problem

• psychological profile
• counseling speech
• hallucination
• long-context forgetting
• evidence grounding

Innovation

• StreamProfile
• Hierarchical Evidence Memory
• Chain-of-Thought Reasoning
• Psychological Profile Generation
• Hallucination Mitigation