MentalArena: Self-play Training of Language Models for Diagnosis and Treatment of Mental Health Disorders

TL;DR

MentalArena tackles the scarcity and privacy concerns surrounding personalized mental health data by introducing a self-play framework where a single model acts as both patient and therapist. The system leverages a Symptom Encoder to simulate patients and a Symptom Decoder to calibrate and refine therapist–patient dialogues through a cognitive bias check, with a Model Optimizer that tunes the model on generated, diagnosis/treatment/medication data. Empirical results across eight benchmarks show that fine-tuned models based on GPT-3.5-turbo and Llama-3-8b outperform strong baselines, including GPT-4o, with notable gains and robust generalization to other medical domains. The approach demonstrates potential for scalable, privacy-conscious personalized care, while analyses address data validity, bias, and forgetting, outlining promising directions for future work.

Abstract

Mental health disorders are one of the most serious diseases in the world. Most people with such a disease lack access to adequate care, which highlights the importance of training models for the diagnosis and treatment of mental health disorders. However, in the mental health domain, privacy concerns limit the accessibility of personalized treatment data, making it challenging to build powerful models. In this paper, we introduce MentalArena, a self-play framework to train language models by generating domain-specific personalized data, where we obtain a better model capable of making a personalized diagnosis and treatment (as a therapist) and providing information (as a patient). To accurately model human-like mental health patients, we devise Symptom Encoder, which simulates a real patient from both cognition and behavior perspectives. To address intent bias during patient-therapist interactions, we propose Symptom Decoder to compare diagnosed symptoms with encoded symptoms, and dynamically manage the dialogue between patient and therapist according to the identified deviations. We evaluated MentalArena against 6 benchmarks, including biomedicalQA and mental health tasks, compared to 6 advanced models. Our models, fine-tuned on both GPT-3.5 and Llama-3-8b, significantly outperform their counterparts, including GPT-4o. We hope that our work can inspire future research on personalized care. Code is available in https://github.com/Scarelette/MentalArena/tree/main

Figures (18)

• Figure 1: MentalArena is a self-play framework for assistance in the diagnosis and treatment of mental health disorder consisting of three modules: Symptom Encoder, Symptom Decoder, and Model Optimizer.
• Figure 2: Case study on pre-training and post-training.
• Figure 3: Symptom Decoder simulates the diagnostic and therapeutic interactions between patients and therapists, enabling the generation of personalized dialogues while addressing cognitive bias.
• Figure 4: Ablation study. Each bar represents the performance of the models trained under different settings.
• Figure 5: Effectiveness analysis of MentalArena.