Skip to main content

Introduction

LLMind: AI-Powered Diagnostic Support for Mental Health Professionals

LLMind is an AI-driven tool with a user-friendly interface designed to assist mental health professionals in the diagnostic process.

How LLMind Works​

LLMind Chat leverages a Retrieval-Augmented Generation (RAG) model based on Gemma 2 (27B parameters), specifically adapted for the ICD-11. This approach enhances its capabilities by:

  • Integrating the strengths of Gemma 2 with a vast ICD-11 knowledge base
  • Accessing and processing relevant information in real-time from the classification manual

Accuracy & Validation​

The diagnostic accuracy of LLMind was rigorously evaluated against the DSM-5-TR Clinical Cases manual, using:

  • Automated metrics for performance assessment
  • Expert validation by mental health professionals

The results suggest that LLMind Chat serves as a reliable decision-support tool, improving diagnostic reasoning and potentially reducing misclassifications.

Core Technologies​

LLMind processes medical case data using:

  • Large Language Models (LLMs)
  • LangChain for RAG-based retrieval
  • ICD-11 codes for diagnosis

By combining these technologies, LLMind enhances clinical decision-making, ensuring more accurate and context-aware diagnoses.