AI agent skill · by @huang-sir1

Radiology Grant

Reframe and polish imaging-AI / radiomics / radiogenomics research as a research grant proposal — convert paper-style "we built a model" into grant logic (clinical need → scientific question → hypothesis → specific aims/研究内容 → technical route/技术路线 → innovation/创新点 → feasibility/可行性 → expected outcomes), and strengthen title, abstract, background/立项依据, aims, key scientific question/关键科学问题, and significance. Primary and most-developed track is NSFC (国家自然科学基金: 青年/面上/地区) and provincial funds (省自然); also covers reframing the same proposal for international funders (NIH R01, ERC Starting/Consolidator/Advanced, Wellcome Trust) including realistic eligibility for China-based applicants and cross-border routes (NSFC international collaboration line, RGC, MSCA, foundations). Use when the user mentions 国自然/省自然/基金申请/标书/立项依据/科学问题/技术路线/创新点/可行性, NIH/R01/ERC/Wellcome/国际基金, or wants to turn a study into a fundable proposal for any of these. Flags weak innovation, thin preliminary data, open-loop technical routes, over-promising, and funder-eligibility mismatches. Reminds the author to verify the current official guidelines; never fabricates preliminary results, citations, or eligibility.

Archived 2026-10-23 · Source updated 2026-07-27 · License: MIT. Check the original repository for the latest version.

SKILL.md instructions

# Research Grant Proposals (国自然 / 省自然 / institutional / international)

Use this skill to turn an imaging-research idea or finished study into a **fundable proposal**.
A grant is not a paper: reviewers fund a **scientific question** and a credible plan to answer
it, not a model. This skill rebuilds the logic and polishes each section to the structure
funders expect. **NSFC/省自然 is the primary, most fully worked-out track**; the same reframing
logic extends to international funders (NIH, ERC, Wellcome) when the team is also pursuing
cross-border funding — see `references/international-grants.md`.

## Core stance

- **Grant logic, not paper logic.** Lead with the clinical need and the **scientific question**,
  then hypothesis → aims → technical route → innovation → feasibility → expected outcomes.
  "Construct a model / improve accuracy" is an engineering task, not a science question — reframe
  it into the mechanism or generalisable principle being tested.
- **Eligibility before drafting, for any international target.** Confirm the author/institution
  can actually apply as lead PI before writing a word — NIH, ERC, and Wellcome each have real,
  currently-changing eligibility constraints that are easy to miss (→
  `references/international-grants.md`). Wasting effort on an inaccessible mechanism is worse
  than a hard "not this track."
- **The key scientific question (关键科学问题) is the spine.** One sharp, answerable question that
  the aims serve. If the aims don't all serve it, the proposal is unfocused.
- **Close the loop.** The technical route (技术路线) must connect need → question → each aim →
  method → expected result → back to the question. Reviewers reject open-loop routes.
- **Innovation must be specific and defensible.** Name the increment (question / data / method /
  validation / mechanism). Avoid "first/领先" without grounding.
- **Feasibility is shown, not asserted.** Preliminary data, team capability, data access, and
  ethics make it credible — never fabricate preliminary results.
- **Don't over-promise.** Aims must be achievable in the period and budget; over-scoping reads as
  naïveté.
- **Verify the current guidelines.** Word limits, format, attachments, ethics, and 限项 rules
  change yearly — confirm against the **current** official 申报指南 (the author must check; this
  skill flags it, it does not have the current year's rules memorised).

## When to use

- "把我的研究改写成国自然/省自然标书 / 帮我写立项依据、科学问题、技术路线、创新点。"
- "Turn this finished study into a grant proposal."
- "Is my innovation point / feasibility strong enough? polish my aims."
- "Reframe 'build a model' into a fundable scientific question."
- "我们也想申请 NIH / ERC / Wellcome，能不能用同一份研究改写？" / "Can this also become an NIH R01 /
  ERC / Wellcome proposal, and am I even eligible to apply?"

## When to open extra files

| File | Open when |
|---|---|
| [references/grant-architecture.md](references/grant-architecture.md) | The section-by-section structure and what each must accomplish (NSFC/provincial) |
| [references/reframe-and-innovation.md](references/reframe-and-innovation.md) | Converting paper/engineering framing into a scientific question + a defensible innovation point |
| [references/feasibility-and-pitfalls.md](references/feasibility-and-pitfalls.md) | Showing feasibility (preliminary data, route closure, team/data/ethics) and the common rejection reasons |
| [references/international-grants.md](references/international-grants.md) | Target is also/instead NIH, ERC, or Wellcome — eligibility reality for China-based applicants, section-mapping from NSFC structure, and more directly reachable cross-border routes |

## Workflow

0. **Identify the target funder(s)** — NSFC/省自然 (default), and/or NIH/ERC/Wellcome/other
   international. For any international target, **check eligibility first**
   (international-grants.md) before investing drafting effort.
1. **Extract the science.** From the idea/study, find the clinical need and the **one scientific
   question** worth funding (reframe engineering goals into mechanism/generalisable principle).
   This is funder-independent — do it once, reuse for every target.
2. **Set the architecture** (grant-architecture.md for NSFC/provincial; international-grants.md
   for the funder-specific section mapping) — title, abstract, 立项依据, 研究目标, 研究内容
   (aims), 关键科学问题, 技术路线, 创新点, 可行性, 预期成果, plan.
3. **Build aims that serve the question** — each aim a testable sub-question with a method and an
   expected result; aims are coherent, not a feature list.
4. **Forge the innovation point** (reframe-and-innovation.md) — specific, defensible, tied to the
   gap; classify the kind of innovation.
5. **Close the technical route** — a loop diagram in prose: need → question → aims → methods →
   expected results → question; mark validation and risk mitigations.
6. **Evidence feasibility** (feasibility-and-pitfalls.md) — preliminary data (real only), team,
   data access, ethics; pre-empt the common rejection reasons.
7. **Polish & guideline-check** — tighten each section; flag every place the author must verify
   the current official 申报指南 (limits, format, 限项, ethics).

## Output contract

0. **`Target funder(s) & eligibility`** (only when international) — funder, whether the
   author/institution can be lead PI under current rules, and the realistic route if not
   (co-PI, alternate mechanism) — before any section drafting.
1. **`Scientific question`** — the one fundable question + hypothesis.
2. **`Section drafts`** — title, abstract, 立项依据, 研究目标/内容, 关键科学问题, 技术路线, 创新点,
   可行性, 预期成果 — drafted or restructured.
3. **`Innovation point`** — specific, classified, defensible.
4. **`Technical-route check`** — is the loop closed? gaps marked.
5. **`Feasibility & risk`** — what supports feasibility; risks + mitigations.
6. **`Weaknesses & fixes`** — thin innovation, weak preliminary data, over-scoping, open loop.
7. **`待核验（中文）`** — current-guideline items the author must confirm (字数/格式/附件/伦理/限项).

## Quality bar

A good proposal makes a reviewer see one sharp scientific question, aims that all serve it, a
closed technical route, a specific innovation, and credible feasibility — written to the funder's
structure, with no fabricated preliminary data and an explicit reminder to verify the current
year's guidelines.

## Handoffs

- Study/validation design behind an aim → `radiology-design`.
- Frontier framing & evidence for the gap → `radiology-frontier` (verify live → `radiology-search`).
- Statistics/sample-size for an aim → `radiology-stats`.
- Ethics/feasibility of data → `radiology-ethics` / `radiology-data`.
- English polish of an English-language proposal → `radiology-polishing`.
- Verifying current call deadlines, page limits, and funder eligibility rules live →
  `radiology-search`.
- This skill drafts and critiques proposals; it does not guarantee funding, does not confirm
  eligibility on the author's behalf, and does not replace the current official 申报指南 or the
  target institution's sponsored-programs/grants office.