🔗 构建健康连接的智能底座
An open-source framework for multi-agent health management and privacy-preserving data intelligence.
OpenLinkage 是由 Linkage App 团队开源的健康智能体框架。 它为开发者提供了一套可扩展的底层架构,用于构建 AI 健康助手、医生辅助分析与个性化健康管理系统。
主要特性:
- 🧠 多智能体协作系统(Multi-Agent Framework)
- 💾 AI 记忆引擎(短期 / 长期)
- 🔒 隐私保护与本地数据计算
- 📊 健康数据融合与趋势建模
- 🔌 开放 API 接口,轻松对接医院、健康设备、运动 App
User
└── Linkage Core
├── Health Butler Agent
├── Nutrition Agent
├── Exercise Agent
├── Medication Agent
├── Security Agent
└── Memory Engine + Privacy Guard
- Backend: Python / FastAPI / LangGraph
- AI Integration: OpenAI GPT / Qwen / Claude
- Vector DB: pgvector / Qdrant
- Privacy: Decentralized encryption inspired by Second-Me
- Optional Frontend: Flutter / React Native
本仓库内包含一个可直接运行的 FastAPI 多智能体演示服务,附带基础校验、概览汇总与单元测试用例,便于快速扩展。
git clone https://github.com/your-org/OpenLinkage.git
cd OpenLinkage
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python run.py
# 运行单元测试(可选,如缺少依赖将自动跳过)
python -m unittest discover -s tests -v启动后,打开 http://localhost:8000/docs 查看 API 文档。
调用 /analyze 接口,模拟健康管家、营养、运动与用药安全智能体协作:
curl -X POST http://localhost:8000/analyze \
-H "Content-Type: application/json" \
-d '{
"user_id": "demo-user",
"symptoms": ["fatigue"],
"goals": ["weight management"],
"lifestyle_notes": "prefers evening workouts"
}'响应会返回每个智能体的建议,便于前端或下游系统进一步处理。
返回结构示例(节选):
{
"user_id": "demo-user",
"overall_summary": "HealthButlerAgent: Follow a consistent sleep schedule to support hormone balance. | NutritionAgent: Prioritize vegetables and lean protein in daily meals.",
"warnings": [
"Chest pain requires immediate evaluation. If severe, call emergency services."
],
"responses": [
{"agent": "HealthButlerAgent", "summary": "...", "recommendations": ["..."]},
{"agent": "NutritionAgent", "summary": "...", "recommendations": ["..."]},
{"agent": "ExerciseAgent", "summary": "...", "recommendations": ["..."]},
{"agent": "MedicationAgent", "summary": "...", "recommendations": ["..."]}
]
}本项目遵循 Apache 2.0 开源协议。 欢迎贡献代码、提出改进意见或提交 Issue!
© 2025 Linkage Team. All rights reserved.