MindGuard: Towards Accessible and Sitgma-free Mental Health First Aid via Edge LLM

Sijie Ji,Xinzhe Zheng,Jiawei Sun,Renqi Chen,Wei Gao,Mani Srivastava

2024-09-16

Abstract:Mental health disorders are among the most prevalent diseases worldwide, affecting nearly one in four people. Despite their widespread impact, the intervention rate remains below 25%, largely due to the significant cooperation required from patients for both diagnosis and intervention. The core issue behind this low treatment rate is stigma, which discourages over half of those affected from seeking help. This paper presents MindGuard, an accessible, stigma-free, and professional mobile mental healthcare system designed to provide mental health first aid. The heart of MindGuard is an innovative edge LLM, equipped with professional mental health knowledge, that seamlessly integrates objective mobile sensor data with subjective Ecological Momentary Assessment records to deliver personalized screening and intervention conversations. We conduct a broad evaluation of MindGuard using open datasets spanning four years and real-world deployment across various mobile devices involving 20 subjects for two weeks. Remarkably, MindGuard achieves results comparable to GPT-4 and outperforms its counterpart with more than 10 times the model size. We believe that MindGuard paves the way for mobile LLM applications, potentially revolutionizing mental healthcare practices by substituting self-reporting and intervention conversations with passive, integrated monitoring within daily life, thus ensuring accessible and stigma-free mental health support.

Computation and Language,Artificial Intelligence,Human-Computer Interaction

What problem does this paper attempt to address?

### The Problem This Paper Attempts to Solve

This paper aims to address the issue of low treatment rates in the diagnosis and intervention of mental health disorders (such as anxiety, depression, etc.), particularly the reluctance of patients to seek help due to social stigma. Specifically:

1. **Prevalence of Mental Health Issues**: Approximately one-quarter of the global population is affected by mental health issues, but less than 25% receive treatment.
2. **Impact of Social Stigma**: More than half of the affected population is unwilling to seek help due to stigma, resulting in extremely low treatment rates.
3. **Limitations of Existing Tools**: Current online tools typically offer only closed-ended questions, lacking personalized interaction, continuous monitoring, and intervention capabilities.

To address these issues, the paper proposes the MindGuard system, a mobile mental health care system driven by a large language model (LLM) based on edge computing. The main features of MindGuard include:
- **Integration of Objective Sensor Data and Subjective Conversational Data**: By combining smartphone sensor data with ecological momentary assessment records to provide personalized screening and intervention conversations.
- **Reduction of Hallucination Phenomena**: Utilizing high-quality datasets, continuous fine-tuning techniques, and other strategies to improve the accuracy and reliability of LLMs in the mental health domain.
- **Real-time Monitoring and Early Intervention**: Continuously monitoring user behavior data to timely identify potential mental health risks and provide personalized intervention suggestions.
- **Privacy Protection**: Running MindGuard on local devices through knowledge distillation and quantization techniques to ensure the security of user data.

Through these designs, MindGuard is expected to become a comprehensive, stigma-free mental health support solution.