MentalArena: Self-play Training of Language Models for Diagnosis and Treatment of Mental Health Disorders
Cheng Li, May Fung, Qingyun Wang, Chi Han, Manling Li, Jindong Wang, Heng Ji
2024-10-10
Summary
This paper introduces MentalArena, a self-play training framework for language models that helps diagnose and treat mental health disorders by simulating patient-therapist interactions.
What's the problem?
Mental health disorders are serious and widespread, but many people do not have access to proper care. One of the challenges in developing effective AI tools for mental health is the lack of personalized treatment data due to privacy concerns. This makes it hard to train models that can provide accurate diagnoses and treatment recommendations.
What's the solution?
To tackle this issue, the authors created MentalArena, which allows language models to train themselves by generating their own personalized data. It includes two key components: the Symptom Encoder, which simulates a patient’s symptoms and behavior, and the Symptom Decoder, which manages the conversation between the patient and therapist. This setup enables the model to learn from its interactions as both a therapist and a patient, improving its ability to provide tailored mental health support.
Why it matters?
This research is important because it could lead to better AI tools for diagnosing and treating mental health issues, potentially increasing access to care for those who need it most. By using self-play to generate training data, MentalArena can help create more effective and personalized mental health solutions while addressing privacy concerns.
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