← Back to Skills Marketplace
unisound-llm

unisound-referral-guidance

by Unisound-LLM · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ pending
81
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install unisound-referral-guidance
Description
基层转诊指导。输入患者病情摘要,判断是否需要转诊、紧急程度、目标科室/医院级别,给出转诊前处置和随附材料清单(JSON + 自然语言摘要)。
README (SKILL.md)

基层转诊指导

概述

面向社区诊所/基层卫生机构医生,给定患者病情摘要(症状、诊断、处理情况),本技能会:

  • 判断是否需要转诊及紧急程度(立即/较急/择期/不需要)
  • 推荐目标转诊科室和医院级别
  • 列出满足的转诊指征(依据国家分级诊疗政策)
  • 给出转诊前必要处置措施
  • 列出应随患者携带的材料

数据安全、隐私与伦理声明

  • 最小必要原则:仅处理转诊决策所必需的病情信息;不要求包含身份信息。
  • 严格脱敏:发送前对可识别身份信息进行脱敏处理。
  • 不做本地持久化:仅在内存中短暂处理;本次调用结束即销毁
  • 医疗边界:本技能输出为转诊决策辅助,不替代临床判断;紧急情况下请优先处置患者。

输入格式

纯文本(UTF-8),包含患者病情关键信息,例如:

患者:男,72岁
主要诊断:急性胸痛,持续20分钟
心电图:II、III、aVF导联ST段压低0.2mV
当前处理:硝酸甘油舌下含服1片,症状无缓解
血压:90/60 mmHg,心率:108次/分
既往史:高血压、糖尿病

也支持 JSON 格式(包含 text/content/case_summary 字段的对象)。

快速开始

# 从 skills 目录运行
python3 clinic/primary-care-assist/referral-guidance/scripts/run.py \
  --input data/clinic-referral/case-001.txt \
  --appkey \x3Cyour-appkey>

# 保存输出到文件
python3 clinic/primary-care-assist/referral-guidance/scripts/run.py \
  --input data/clinic-referral/case-001.txt \
  --appkey \x3Cyour-appkey> \
  --output runs/clinic-referral/case-001.json

参数说明

  • --input PATH必填。患者病情摘要文件路径(txt 或 json,UTF-8)。
  • --appkey STRING必填。调用内部医疗大模型的鉴权 key,由平台分配。
  • --output PATH:输出文件路径(默认:打印到 stdout)。
  • --base URL:内部大模型 base URL(默认:https://maas-api.hivoice.cn/v1)。
  • --model STRING:模型名称(默认:u1-insuremed)。
  • --timeout SECONDS:HTTP 超时秒数;0 表示一直等待(默认:0)。
  • --encoding STRING:输入文件编码(默认:utf-8)。

输出约定

输出分两部分:

JSON 结构

{
  "referral_decision": "建议立即转诊",
  "urgency": "紧急(立即)",
  "target_department": "心内科/急诊科",
  "target_hospital_level": "三级医院",
  "referral_reason": "急性下壁心肌缺血,血压低,硝酸甘油无效,疑似急性冠脉综合征",
  "referral_criteria_met": [
    "心电图提示ST段改变",
    "血流动力学不稳定(血压90/60mmHg)",
    "胸痛持续且治疗无效"
  ],
  "pre_transfer_measures": [
    "建立静脉通道",
    "持续心电监护",
    "给予阿司匹林300mg嚼服(若无禁忌)",
    "拨打120急救"
  ],
  "transfer_documents_required": [
    "转诊单",
    "心电图原件/照片",
    "用药记录"
  ],
  "can_manage_locally_if_no_referral": null
}

自然语言摘要:以"【摘要】"开头,说明转诊建议和关键注意事项。

依赖

运行环境

  • Python 3.7+(仅使用标准库,无需额外安装)

外部 API

  • 内部医疗大模型:https://maas-api.hivoice.cn/v1/chat/completions

备注

  • 转诊标准参考国家分级诊疗相关政策及常见疾病转诊指南
  • 紧急情况下,系统会同时在 pre_transfer_measures 中列出院前急救措施
  • 发布约束:示例输入、运行输出均放在 skill 包外(data/runs/),skill 目录内仅保留可发布的核心文件
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install unisound-referral-guidance
  3. After installation, invoke the skill by name or use /unisound-referral-guidance
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
clinic-referral-guidance v1.0.0 - Initial release of the clinic-referral-guidance skill for primary care doctors. - Given a patient case summary, determines transfer need, urgency, target department/hospital level, and provides referral criteria, pre-transfer measures, and checklist of accompanying documents. - Supports both plain text and specific JSON input formats. - Outputs both structured JSON and a natural-language summary. - Includes clear privacy, security, and usage guidelines. - Provides command-line usage instructions and parameter details for quick start.
Metadata
Slug unisound-referral-guidance
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is unisound-referral-guidance?

基层转诊指导。输入患者病情摘要,判断是否需要转诊、紧急程度、目标科室/医院级别,给出转诊前处置和随附材料清单(JSON + 自然语言摘要)。 It is an AI Agent Skill for Claude Code / OpenClaw, with 81 downloads so far.

How do I install unisound-referral-guidance?

Run "/install unisound-referral-guidance" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is unisound-referral-guidance free?

Yes, unisound-referral-guidance is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does unisound-referral-guidance support?

unisound-referral-guidance is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created unisound-referral-guidance?

It is built and maintained by Unisound-LLM (@unisound-llm); the current version is v1.0.0.

💬 Comments