deepevidence

deepevidence循證醫學AI助手

專業領域v1.0.15
作者:deepevidence更新於 2026-09-11共 16 個版本 · 最新版本釋出於 2026-09-11deepevidence@clawhub_cindy8753/deepevidence
醫療健康#evidence#guideline#drug#medical-literature#clinical-decision

據 AlphaMoat 統計,截至 2026-09,deepevidence循證醫學AI助手 累計下載 1,860 次、獲 1 個星標,在全部 5,078 個 AI 技能中排名第 4,529 位,在「專業領域」分類(465 個)中排名第 401 位。

累計安裝
187
累計下載
1860
星標
1
全站排名
#4,529
全站前 90% · 「專業領域」前 87%
平台累計安裝累計下載星標評論
ClawHub ↗27129611
SkillHub ↗18718601—

各平台的計數口徑不同,統計時間不一致。

這個技能是做什麼的

主文涵蓋指南解讀、藥物安全與相互作用、劑量調整、臨床研究證據和醫學文獻等問題,並強調回答應作為臨床參考,而非診斷或治療指令。它不適用於急診分診或替代執業臨床人員;處理患者可識別資訊還需要相應授權、同意和合規措施。

資料釋出於 2026-10-27

可以用來做什麼

  • 整理指南、藥物安全和臨床研究證據
  • 查詢有文獻依據的臨床參考內容

什麼時候用

  • 醫生需要查閱醫學文獻或指南支援的參考資訊時
  • 涉及急症或患者識別資訊時,先遵守主文的轉介與隱私邊界

示例提示詞

以下是根據技能文件生成的示例,不代表實際執行結果。

總結某臨床問題可參考的指南與研究證據,並保留引用
說明此問題需要臨床參考還是應立即轉交急診處理

月度資料趨勢

累計下載1860
279112819772026-03 · 累計下載 3962026-05 · 累計下載 8972026-06 · 累計下載 7432026-07 · 累計下載 12652026-08 · 累計下載 13962026-09 · 累計下載 186026-0326-09
星標1
0122026-03 · 星標 12026-05 · 星標 12026-06 · 星標 12026-07 · 星標 12026-08 · 星標 12026-09 · 星標 126-0326-09

2026-09 起統計方法調整:同名技能按作者分別統計,並覆蓋全部技能。圖中虛線為調整前的月份,與之後的資料不直接比較,也不計算跨越調整的增長。

月份累計下載環比星標
2026-091860—1
2026-081396+10.4%1
2026-071265+70.3%1
2026-06743-17.2%1
2026-05897—1
2026-03396—1

SKILL.md 原文

以下是作者釋出的原始檔案,按純文字顯示,本站未做修改或稽核;其中的連結不可點選,命令請核實後再執行。抓取於 2026-10-23。

3.2 KB · ClawHub v1.0.15
---
name: deepevidence-api
description: >
  Use DeepEvidence for evidence-based clinical questions, including guideline
  interpretation, drug safety and interactions, dose adjustment, clinical-trial
  evidence, medical literature, and supported medical-image analysis.
license: Proprietary
---

# DeepEvidence
展開全部(共 72 行)

## Purpose

Use this skill to call DeepEvidence for physicians' evidence-based clinical decision support. DeepEvidence returns clinical reference content grounded in retrieved medical literature and guidelines.

## When To Use

Use this skill when the user needs literature-backed clinical reference, such as:

- guideline interpretation
- drug safety, contraindications, interactions, or dose adjustment
- clinical trial or study evidence synthesis
- medical literature questions
- supported medical image analysis where authorized

## When Not To Use

Do not use this skill as a general chatbot, emergency triage tool, first-aid replacement, or substitute for a licensed clinician.

If the user describes urgent symptoms such as chest pain, suspected stroke, trouble breathing, altered consciousness, poisoning, overdose, severe allergic reaction, uncontrolled bleeding, infant seizures, severe dehydration, or high fever with mental status changes, advise immediate local emergency care.

## Workflow

1. Confirm `DEEPEVIDENCE_API_KEY` is available in the runtime environment.
2. Confirm Python 3 and network access are available.
3. Avoid sending patient-identifiable information unless authorization, consent, contract, and compliance controls are in place.
4. Call the canonical script:

```bash
python scripts/chat.py --query "In T2D with CKD, how should metformin dose be adjusted by eGFR?"
```

5. For authorized image input, pass one or more image URLs:

```bash
python scripts/chat.py --query "请分析这张医学图片的关键信息。" --image-url "https://example.com/medical-image.jpg"
```

6. Present the answer as clinical reference, not as diagnosis or treatment instruction.

## Hard Rules

- Use `DEEPEVIDENCE_API_KEY` only from environment variables; never ask users to paste keys into chat.
- Do not print, log, screenshot, commit, or package API keys.
- Use `DEEPEVIDENCE_USER_ID` or `--user` only for stable opaque non-PII identifiers.
- Preserve citation markers returned by DeepEvidence; do not renumber, remove, reorder, or invent citations.
- Do not invent DOI, URL, journal, author, year, or other reference metadata.
- If references are missing or incomplete, say so explicitly.
- Do not interpret empty, timed-out, or failed retrieval as "no risk" or "no evidence".
- If evidence content was successfully retrieved from DeepEvidence, end the response with:

```text
> Source: DeepEvidence
```

## Reference Routing

- For endpoint, request schema, image content parts, streaming, and public API limits: read [references/api.md](references/api.md).
- For medical safety, emergency boundaries, PII, patient data, and citation integrity: read [references/safety.md](references/safety.md).
- For status-code handling, timeout, retry, invalid response, and troubleshooting: read [references/errors.md](references/errors.md).
- For curl, Python, Node, and system integration examples: read [references/integration.md](references/integration.md).

「專業領域」分類下的熱門技能

資料月份:2026-09 · 下載/星標/安裝數聚合自公開技能註冊庫(ClawHub、SkillHub),按月更新。口徑說明見資料方法。