grant-proposal-assistant（拨款提案助理）

Grant Proposal Assistant

Table of Contents

• Core Questions
• Workflow
• Section Frameworks
• Reviewer Mindset
• Guardrails
• Quick Reference

Related skills (not this one):

• Manuscripts: scientific-manuscript-review
• Fellowship personal statements: career-document-architect
• Letters of recommendation: academic-letter-architect

Core Questions

Every grant proposal must convincingly answer these four questions:

1. What is the central hypothesis?

• Testable, specific, falsifiable
• Not just "we will study X" but "we hypothesize that X causes Y through mechanism Z"

2. Why is the problem important NOW?

• What gap exists in current knowledge?
• Why is this gap significant for the field/patients/society?
• Why is this the right time (new tools, preliminary data, shifting paradigm)?

3. What makes the approach innovative?

• What is genuinely new (concept, method, application)?
• How does this advance beyond incremental improvement?
• Innovation in approach AND/OR innovation in what will be learned

4. Is the plan feasible and logical?

• Can this team do this work in this timeframe with these resources?
• Do aims build logically without fatal dependencies?
• Are pitfalls anticipated with alternatives ready?

Workflow

Copy this checklist and track your progress:

Grant Proposal Progress:
- [ ] Step 1: Identify grant mechanism and constraints
- [ ] Step 2: Core questions audit
- [ ] Step 3: Specific Aims review (1-page)
- [ ] Step 4: Significance section review
- [ ] Step 5: Innovation section review
- [ ] Step 6: Approach section review (per aim)
- [ ] Step 7: Reviewer alignment check
- [ ] Step 8: Compliance verification

Step 1: Identify Grant Mechanism and Constraints

Determine mechanism (R01, R21, K, NSF, Foundation). Note page limits, required sections, and review criteria. R01 = 12 pages; R21 = 6 pages; K = 12 pages + career development. See resources/methodology.md for mechanism-specific guidance.

Step 2: Core Questions Audit

Read entire proposal looking ONLY for answers to the four core questions. Mark where each is addressed (or missing). Flag unclear hypotheses, weak significance, or missing innovation. See resources/methodology.md for audit checklist.

Step 3: Specific Aims Review

Evaluate the 1-page Aims against the gold standard: Opening hook → Gap → Hypothesis → Aims (testable, independent, coherent) → Impact. This is the most important page. See resources/template.md for structure.

Step 4: Significance Section Review

Check: What is the problem? Why does it matter? What will change if successful? Look for explicit gap statements and impact predictions. See resources/methodology.md for evaluation criteria.

Step 5: Innovation Section Review

Check: What is genuinely new? Be specific (not "innovative approach" but "first application of X to Y"). Innovation can be conceptual, methodological, or in expected outcomes. See resources/methodology.md for evaluation criteria.

Step 6: Approach Section Review

For EACH aim: Rationale (why this aim?) → Strategy (how?) → Expected outcomes → Pitfalls → Alternatives. Check for adequate controls, statistical power, timeline realism. See resources/template.md for per-aim structure.

Step 7: Reviewer Alignment Check

Read as a non-expert reviewer would. Can they understand significance without deep domain knowledge? Are impact statements prominent? Is the writing accessible? See resources/methodology.md for reviewer simulation.

Step 8: Compliance Verification

Check page limits, required sections, biosketch format, reference formatting. Verify all required components present. Validate using resources/evaluators/rubric_grant_proposal.json. Minimum standard: Average score ≥ 3.5.

Section Frameworks

Specific Aims Page (1 page)

The most important page of your grant.

Structure:

OPENING PARAGRAPH (4-6 sentences)
- Hook: Why this problem matters (significance)
- Gap: What's missing in current understanding
- Long-term goal: Your program of research
- Central hypothesis: Testable, specific
- Rationale: Why this hypothesis is reasonable (preliminary data)

AIM 1: [Verb phrase describing objective]
- Brief description (2-3 sentences)
- Expected outcome and interpretation
- Must be testable and achievable

AIM 2: [Verb phrase describing objective]
- Brief description (2-3 sentences)
- Expected outcome and interpretation
- Independent of Aim 1 (can proceed if Aim 1 fails)

AIM 3 (optional): [Verb phrase describing objective]
- Brief description (2-3 sentences)
- May integrate findings from Aims 1-2

CLOSING PARAGRAPH (2-3 sentences)
- Expected outcomes of the project
- Impact: How this advances the field
- Future directions this enables

Significance Section

Goal: Convince reviewers the problem matters

Key elements:

1. The Problem: What clinical/scientific problem exists?
2. Current State: What's known, what's been tried?
3. The Gap: What critical question remains unanswered?
4. Impact of Gap: What's the cost of not knowing?
5. If Successful: What changes? Be specific.

Red flags:

• ❌ Generic statements ("cancer is bad")
• ❌ No clear gap statement
• ❌ Impact statements too vague ("will advance the field")
• ✅ Specific gap, specific impact, quantifiable where possible

Innovation Section

Goal: Show this is not incremental

Types of innovation:

1. Conceptual: New framework, paradigm, or understanding
2. Methodological: New technique, approach, or model
3. Application: Known method applied to new problem
4. Expected Outcomes: Will generate novel insights

Format:

• Use bullet points for scannability
• Start each with "This project is innovative because..."
• Be specific, not vague

Approach Section (Per Aim)

Structure for each aim:

AIM X: [Title]

RATIONALE (1 paragraph)
Why is this aim necessary? How does it address the hypothesis?

PRELIMINARY DATA (if applicable)
What have you already shown that supports feasibility?

STRATEGY (2-4 paragraphs)
- Experimental design
- Methods and procedures
- Controls (positive and negative)
- Statistical analysis plan

EXPECTED OUTCOMES
What results do you expect? How will you interpret them?

POTENTIAL PITFALLS AND ALTERNATIVES
What could go wrong? What's your backup plan?

TIMELINE/MILESTONES
When will this be completed? Dependencies on other aims?

Reviewer Mindset

How Study Sections Work

• Reviewers assigned based on expertise (but may not be YOUR exact field)
• Primary reviewers read carefully; secondary skim
• 3 reviewers score; others may not read deeply
• Scored on: Significance, Investigators, Innovation, Approach, Environment
• Overall Impact = "How important is this research?"

What Reviewers Look For

Good proposals make reviewers' jobs easy:

• Clear hypothesis on page 1
• Explicit significance statements
• Obvious innovation points (bulleted)
• Logical aim flow
• Pitfalls acknowledged with alternatives

Proposals get criticized for:

• Vague hypotheses ("We will explore...")
• Missing controls
• Overly ambitious scope
• Aim dependencies (if Aim 1 fails, whole project fails)
• No preliminary data for risky approaches
• Unclear statistical plans

Guardrails

1. Testable hypothesis: The hypothesis should be falsifiable, not just a goal
2. Explicit gaps: State what is unknown, not just what you will do
3. Real innovation: Be specific, not just "innovative approach"
4. Independent aims: The project should survive if one aim fails
5. Feasibility evidence: Include preliminary data for risky elements
6. Power calculations: Know sample sizes and why they are sufficient
7. Pitfall acknowledgment: Show anticipated problems and alternatives

Common pitfalls:

• ❌ Fishing expedition: "We will determine..." without hypothesis
• ❌ Aim dependency: Aim 2 impossible without Aim 1 success
• ❌ Scope creep: Too ambitious for budget/time
• ❌ Missing controls: Experiments without proper comparisons
• ❌ Vague statistics: "Data will be analyzed appropriately"
• ❌ No alternatives: Assuming everything will work

Quick Reference

Key resources:

• resources/methodology.md: Grant mechanisms, audit checklists, reviewer perspective
• resources/template.md: Specific aims template, approach per-aim structure
• resources/evaluators/rubric_grant_proposal.json: Quality scoring

Page limits:

NIH scoring:

• 1-3: Exceptional to Excellent (funded)
• 4-5: Very Good to Good (may fund)
• 6-7: Satisfactory to Fair (unlikely)
• 8-9: Marginal to Poor (not funded)

Typical writing time:

• Specific Aims (polished): 3-5 days
• Full R01 first draft: 4-6 weeks
• R21 first draft: 2-3 weeks
• Revision cycle: 1-2 weeks per round

Inputs required:

• Research idea with preliminary data
• Grant mechanism and deadline
• Institutional resources available

Outputs produced:

• Structured grant sections
• Commentary on strengths/weaknesses
• Reviewer-perspective critique