人工智能医学实验室预约系统
一个全栈、人工智能驱动的医学实验室测试预约系统，使用 AI 代理, LangGraph编排，以及 基于MCP的工具调用.\ 该系统演示了LLM支持的代理如何安全地推理、调用后端工具和管理现实世界的工作流程，如医疗实验室调度和预订。
______________________________________________________________________
▶️ 如何运行应用程序>>>>演示视频
🎥 请点击此处观看完整演示：\ 👉 https://www.youtube.com/\
演示显示：
- 人工智能聊天机器人预约实验室测试
- 基于工具的推理和槽验证
- 用于管理实验室操作的管理面板
______________________________________________________________________
🚀 如何在本地运行项目
1.️⃣ 克隆存储库
git clone https://github.com/AnjumZahid/agentic-ai-lab-booking.git
cd agentic-ai-lab-booking
2️⃣ Create and activate virtual environment
conda create -n lab_ai python=3.10 -y
conda activate lab_ai
3️⃣ Install dependencies
*Create .env file and put:
GOOGLE_API_KEY = "Your Google API KEY"
4️⃣ Start backend server (FastAPI) & crud_backend.py
python server.py
uvicorn backend.crud_backend:app --reload
Backend will be available at:
http://localhost:8000
5️⃣ Run the Streamlit frontend
streamlit run frontend_run.py OR streamlit run frontend_run.py
Frontend will open in your browser:
http://localhost:8501
🧠 System Architecture (High Level)
AI Chatbot (LangGraph)
Orchestrates multi-step reasoning, tool execution, and conversation state.
MCP Tool Layer
Secure tool calls for schedules, availability, bookings, and validation.
FastAPI Backend
Business logic, booking rules, and database operations.
Streamlit Frontend
User chatbot interface + Admin dashboard.
SQLite Database
Stores tests, schedules, doctors, holidays, and bookings.
🛠 Key Features
🤖 AI Chatbot for Lab Booking
Natural language booking
Availability checks
Doctor requirement handling
Context-aware, streaming responses
🔗 Agentic AI with MCP Tool Calls
LLM decides when and which tool to call
No hardcoded flows
Safe, structured execution
⚙️ Admin Panel
Manage lab tests
Configure schedules & windows
Add doctors & holidays
Assign tests to doctors
View & manage bookings
📅 Advanced Booking Logic
Lab holidays & partial days
Test-specific closures
Doctor availability
Proportional slot calculation
Concurrency-safe booking
🧩 LangGraph Orchestration
Threaded conversations
Stateful agent execution
Reliable tool orchestration
Production-grade control vs no-code tools
🧪 Tech Stack
Python
LangGraph
LangChain
MCP (Model Context Protocol)
FastAPI
Streamlit
SQLite
LLMs (tool-calling capable)
📌 Why LangGraph (vs CrewAI / n8n)
Fine-grained control over execution
Stateful, multi-turn agent flows
Safe tool invocation
Easier debugging & observability
Designed for real production systems
📂 Project Structure
.
├── backend/ # FastAPI backend & booking logic
├── frontend/ # Streamlit admin pages
├── utils/ # Helper & validation utilities
├── langgraph_mcp_backend.py# Agent + tool orchestration
├── frontend_chatpage.py # Chat interface
├── frontend_run.py # Streamlit entry point
├── server.py # FastAPI server
├── requirements.txt
└── README.md
🔒 Disclaimer
This project is a technical demo for showcasing agentic AI and orchestration patterns.
It is not a certified medical system.
🤝 Connect
If you’re interested in:
Agentic AI
LangGraph orchestration
MCP tool calling
Healthcare AI workflows
Feel free to connect on LinkedIn or reach out on GitHub.
⭐ If you find this project useful, please consider starring the repo!