↓Skip to main content
  1. 『Human Health Plan』Project White Paper/

SMD Artificial Intelligence

Smart Diagnosis Vision Approach
Human Health Plan
Author
Human Health Plan
SMD Evidence-Based Medicine: Theory & Application.
Table of Contents

Preface
#

  1. SMD stands for “Systemic Medicine Doctor” and the domain “SM.Doctor”.
  2. SMD Core Philosophy: First principles of health, health systems engineering, and integrated healing through healthcare and wellness.
  3. SMD Therapy Characteristics: Organic integration of Chinese and Western medical therapies, psychological therapy, and natural therapies.
  4. SMD Evidence-Based Medicine: A closed-loop medical system integrating SMD-Methodology-Driven case analysis, treatment planning, and outcome validation.
  5. Human Health Project: A comprehensive health initiative based on SMD methodology.

SMD-AI Construction Plan
#

Building the SMD-AI Intelligent Diagnosis and Treatment System (including generative AI and wearable health monitoring devices) based on the DeepSeek-v large language model, RAG (Retrieval-Augmented Generation) technology, MCP protocol, knowledge base, and SMD case database.

I. Technical Introduction
#

  1. SMD-LLM (Large Language Model)
  • Core Capabilities: Enables language understanding, generation, and logical reasoning through pre-trained models.
  • Key Challenges: Knowledge cutoff (dependent on training data timeframe), factual errors, lack of real-time information, and insufficient domain-specific expertise.
  1. SMD-RAG (Retrieval-Augmented Generation)
  • Core Capabilities: Enhances output accuracy and relevance by retrieving and integrating external knowledge (structured/unstructured data) into the LLM generation process.
  • Technical Approach:
    • Retrieval Module: Recalls candidate documents from knowledge bases using vector similarity (e.g., embeddings) or semantic matching (e.g., BM25).
    • Generation Module: LLM synthesizes retrieved information to generate responses with controlled formatting (e.g., source citations).
  1. SMD-Knowledge Base
  • Core Capabilities: Stores domain-specific structured/semi-structured data.
  • Role: Provides reliable knowledge for RAG, addressing LLM limitations like outdated information and lack of domain expertise.
  1. SMD-Case Database
  • Core Capabilities: Stores structured/semi-structured case data on disease rehabilitation.
  • Role: Supplies evidence-based treatment cases to improve diagnostic precision.

II. Application Scenarios
#

  • Scenario: Precision Q&A Diagnosis System.
  • Workflow: User query → RAG retrieves authoritative documents from knowledge/case databases → LLM synthesizes information → Generates sourced answers.
  • Advantage: Avoids LLM hallucinations and enhances response credibility.

III. Synergistic Advantages
#

  1. Accuracy: Validates LLM outputs via knowledge bases to reduce hallucinations.
  2. Timeliness: Integrates real-time data to cover post-training events.
  3. Domain Depth: Compensates for LLM gaps in specialized knowledge.
  4. Controllability: Guides LLM outputs using retrieval results for compliance.

IV. Summary
#

  • SMD-LLM: The “brain” for generative capabilities, requiring external knowledge supplementation.
  • SMD-RAG: The “bridge” connecting LLM and knowledge bases to resolve factual issues.
  • SMD-MCP: Standardizes interactions between AI systems and external data/services.
  • SMD-Knowledge Base: The reliable knowledge foundation for vertical applications.
  • SMD-Case Database: The evidence-based case foundation for precise medical scenarios.

The integration of these components, and future deployment in embodied AI robots, will advance SMD-AI from “general dialogue” to “specialized diagnostic agent.”

V. Demo
#


　 　　Schedule a Demo