DemMA: Dementia Multi-Turn Dialogue Agent with Expert-Guided Reasoning and Action Simulation

TL;DR

DemMA is a novel dementia patient simulation framework that combines expert-guided persona construction, multi-modal behavior modeling, and an efficient reasoning-action generation method, advancing realistic and scalable dementia dialogue simulation.

Contribution

It introduces a comprehensive, expert-informed multi-turn dementia simulation system with integrated nonverbal behavior modeling and a Chain-of-Thought distillation framework for efficient reasoning and action generation.

Findings

DemMA outperforms baseline models in expert and LLM evaluations.

The Chain-of-Thought framework enables joint reasoning and behavior generation in one pass.

Nonverbal behavior modeling enhances simulation realism.

Abstract

Simulating dementia patients with large language models (LLMs) is challenging due to the need to jointly model cognitive impairment, emotional dynamics, and nonverbal behaviors over long conversations. We present DemMA, an expert-guided dementia dialogue agent for high-fidelity multi-turn patient simulation. DemMA constructs clinically grounded dementia personas by integrating pathology information, personality traits, and subtype-specific memory-status personas informed by clinical experts. To move beyond text-only simulation, DemMA explicitly models nonverbal behaviors, including motion, facial expressions, and vocal cues. We further introduce a Chain-of-Thought distillation framework that trains a single LLM to jointly generate reasoning traces, patient utterances, and aligned behavioral actions within one forward pass, enabling efficient deployment without multi-agent inference. Extensive evaluations with experts, medical students, and LLM judges demonstrate that DemMA significantly outperforms strong baselines across multiple metrics.