nfyy-ckd-risk-warning-mcp
This server is a mock MCP for a pediatric CKD risk warning agent, providing simulated tools for demo purposes. You can list 15 fictional patients, filterable by scenario type (normal, abnormal, boundary, exception); fetch biochemical trends as if from a hospital LIS, including current and historical lab results, percent changes, and height/weight Z-scores; evaluate risk using a deterministic hard-rule engine with 13 built-in rules to determine risk level (L0–L3) and triggered rules, supporting either patient ID auto-fetch or manual lab value input to prevent LLM arithmetic hallucinations; trigger a structured CKD-ALERT/1.0 warning event, simulate delivery to a message queue and push channels (doctor workstation, WeChat mini-program), return a delivery receipt, and optionally check and correct agent-suggested risk levels; and retrieve clinical SOPs for specific rules or all rules, with doctor, nurse, and parent perspectives and evidence-based rationales. Data is simulated and not for real clinical use.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@nfyy-ckd-risk-warning-mcp对患儿 C_8801 进行风险研判,今日肌酐 165,血钾 5.8,请给出预警等级和处置建议。"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
童肾营养师·风险预警 Agent —— Mock MCP Server
团队编号:NFYY038 | 项目:南方医科大学南方医院「童肾营养师」多智能体辅助决策系统 子智能体:风险预警 Agent(复赛演示版)
本仓库是「童肾营养师」多智能体系统中风险预警 Agent 的自定义 MCP(Model Context Protocol)服务实现。它解决复赛阶段无法接入真实 HIS / LIS、无法通过小程序真实推送的约束,用模拟数据 + 确定性硬规则引擎 + 北向报文/MQ 模拟,完整复现「风险研判 → 等级判定 → 结构化报文推送 → 临床处置 SOP」的预警能力。
1. 它能做什么
能力 | 说明 |
南向 Mock·LIS 趋势 |
|
硬规则引擎 |
|
北向报文 + MQ 模拟 |
|
临床处置 SOP |
|
内置 5 个 MCP 工具、15 例虚构患儿(覆盖 normal / abnormal / boundary / exception 四类场景)、13 条硬规则。
Related MCP server: groowooth
2. 架构与数据流
医生「研判指令输入」
│
▼
┌──────────────────────────────────────────────────────────┐
│ 风险预警 Agent(8B 模型,仅组织自然语言) │
│ 1. list_mock_patients → 选定患儿 │
│ 2. fetch_patient_biochemical_trends (南向 Mock①/LIS) │
│ 3. evaluate_risk_rules → 硬规则引擎判定等级 │
│ 4. trigger_warning_event (北向 Mock②/推送+MQ) │
│ 5. get_clinical_sop → 处置 SOP │
└──────────────────────────────────────────────────────────┘
│ 结构化 JSON 报文 (CKD-ALERT/1.0)
▼
┌──────────────────────────────────────────────────────────┐
│ 北向接口 / 消息队列(模拟 ⇒ 上线替换为 Kafka / 企微网关) │
│ topic: ckd.alert.pediatric │
└──────────────────────────────────────────────────────────┘
│ 透传
▼
医生工作站待办卡片 / 家长端微信小程序订阅消息硬规则(防幻觉红线)示例:肌酐较基线涨幅 ≥50% → L1 一级 AKI;血钾 >5.5 mmol/L → L1 高钾危急;白蛋白 <35 g/L 且体重 Z 连续下降 → L2 PEW。完整 13 条见 src/nfyy_ckd_risk_warning/rules.py 与 knowledge/。
3. 安装与运行
方式一:PyPI / uvx(魔搭 MCP 广场部署用)
# 以 stdio 方式拉起服务(魔搭/Nexent 自定义 MCP 即如此调用)
uvx nfyy-ckd-risk-warning-mcp服务配置 JSON(魔搭 MCP 广场「自定义创建-可托管部署」填法):
{ "mcpServers": { "nfyy038-risk-warning": { "command": "uvx", "args": ["nfyy-ckd-risk-warning-mcp"] } } }
方式二:源码安装(开发 / 评测)
python -m venv .venv && source .venv/bin/activate
pip install -e .
nfyy-ckd-risk-warning-mcp # 等价于 python -m nfyy_ckd_risk_warning本地 SDK 调用(不依赖 MCP 传输,便于单测/演示)
from nfyy_ckd_risk_warning import server
# 输入今日生化报告 API 新结果,引擎自动对比上一次历史结果并重新研判
out = server.trigger_warning_event("C_8808", today_creatinine=165, today_potassium=5.8, today_albumin=33)
# out["reassessment"] -> 历史 L3 → 本次 L1 对照;out["payload"] / out["markdown"] -> 报文与 SOP4. 复赛 Demo 演示要点
复赛无真实北向接口时,用「研判指令输入 + 今日生化报告 API 模拟** 触发即可证明预警能力:
请对患儿 C_8808 进行风险研判。今日生化报告 API 回传:血肌酐 165 μmol/L,血钾 5.8 mmol/L,血清白蛋白 33 g/L。 (患儿上一次历史结果:肌酐 95、血钾 5.4、白蛋白 43;历史系统判定 L3。)
Agent 经 fetch(历史记录) → trigger(输入今日新报告,引擎自动对比上一次结果并重新研判) 两步,输出:
重新研判对照:历史系统判定 L3 → 本次基于新数据重新研判 L1(肌酐 95→165 +73.7%≥50% → R-01;血钾 5.8>5.5 → R-02;白蛋白 33<35 且体重 Z 连降 → R-04),明确标注「历史等级被新数据推翻」;
推送 JSON 报文(
CKD-ALERT/1.0契约,含等级、命中规则、复查、循证);临床处置 SOP(Markdown:主诊医生 / 护理 / 家长 / 饮食红线 / 复查);
一句话结论摘要(模型组织,附免责声明)。
说明:模拟数据 = 患儿历史记录;今日新血检结果(生化报告 API 模拟)是重新研判的驱动方——新数据优先、不沿用历史等级。本工具不设「以系统/引擎为准」的强制纠偏,历史系统判定仅作对比展示,最终等级由「今日新数据 + 上一次历史基线」的硬规则计算决定。
5. 目录结构
nfyy038-ckd-risk-warning-mcp/
├── src/nfyy_ckd_risk_warning/
│ ├── data/patients.json # 15 例模拟患儿数据集(虚构)
│ ├── rules.py # 硬规则引擎(13 条)
│ ├── sop.py # 临床处置 SOP 文案库 + 循证
│ ├── store.py # 模拟 LIS/HIS 南向数据访问层
│ ├── payload.py # 北向报文 + 消息队列模拟 + Markdown 渲染
│ ├── server.py # FastMCP Server(5 个工具)
│ ├── __main__.py / __init__.py
├── scripts/ # 数据生成 / 测评集生成 / stdio 冒烟
├── prompts/ # 系统提示词(四模块模板)
├── knowledge/ # 知识库 SOP 简化版(iData RAG 挂载参考)
├── datasets/ # 模拟数据集 + 测评数据集 + 数据/测评说明
├── docs/ # 部署/接入/Demo/材料清单
├── tests/test_engine.py # 确定性测试套件(15 例全通过)
└── pyproject.toml # PyPI 发布配置(包名 nfyy-ckd-risk-warning-mcp)6. 测试
pip install -e .
python tests/test_engine.py # 4 项全部通过:15 例预期等级+规则、报文/MQ 契约
python scripts/_smoke_stdio.py # MCP stdio 握手:暴露 5 个工具7. 免责声明
本项目及全部数据均为虚构合成数据,仅用于「南方医科大学南方医院智能体创新大赛」复赛演示,不可用于任何真实临床决策。所有处方级用药调整须由主诊医生决定。
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- Flicense-qualityDmaintenanceA comprehensive MCP server that bridges AI applications with FHIR healthcare data systems, enabling patient data access, clinical data retrieval, and data quality assessment.Last updated4
- Alicense-qualityDmaintenanceMCP server that enables AI agents to assess child growth, plot growth curves, and interpret z-scores using WHO and China NHC standards.Last updatedMIT
- Flicense-qualityBmaintenanceA learning MCP server providing synthetic FHIR patient data with read tools and a gated write workflow (propose → human approve → commit) with structured audit logging.Last updated
- Flicense-qualityBmaintenanceMCP server for infant fever management that tracks symptoms, medication, and lab results, providing guideline-based escalation levels, pharmacokinetic predictions, and structured clinical reports.Last updated
Related MCP Connectors
MCP gateway federating 21 biomedical MCP servers behind one endpoint: gnomAD, ClinVar, HPO, VEP.
MCP server for medicare-coverage
Hosted MCP for denial, prior auth, reimbursement, workflow validation, batch scoring, and feedback.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/rpint/nfyy-ckd-risk-warning-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server