Academic Chapter Writer

Academic Chapter Writer

Transform topics into publishable textbook chapters with comprehensive research, proper citations, and human-sounding academic prose.

Overview

This skill creates full-length academic textbook chapters (5,000-15,000 words) through an agentic workflow:

1. Research - Search PubMed, compile 20-30 key references 2. Plan - Generate detailed TOC with word counts → GET APPROVAL 3. Write - Section-by-section with citations 4. Compile - Full chapter with Vancouver-style reference list

Workflow

Phase 1: Topic Analysis & Research

When user provides topic:

1. Clarify scope:

• Topic: [exact topic]
• Level: undergraduate | graduate | professional | advanced
• Target length: [default 8,000 words]
• Special focus: [any specific angles]

2. Research via PubMed MCP:

   Use PubMed:search_articles for:
   - Main topic + "review" (find reviews)
   - Main topic + "clinical trial" (find trials)
   - Main topic + "guideline" (find guidelines)
   - Main topic + "meta-analysis" (find syntheses)

3. Compile reference library:

• Target 20-30 papers
• Prioritize: NEJM, JAMA, Lancet, JACC, Circulation, EHJ
• Include: 3-5 landmark trials, 2-3 meta-analyses, 1-2 guidelines, reviews

4. Report to user: "I've searched PubMed and compiled [N] key references including:

• [X] landmark trials
• [X] meta-analyses
• [X] guidelines/reviews

Ready to generate table of contents?"

Phase 2: Table of Contents Generation

Generate structured TOC with word allocations:

# [Chapter Title]

## 1. Introduction (~500 words)
   - Opening hook (clinical scenario)
   - Scope and importance
   - Chapter overview

## 2. [Historical Context / Background] (~800 words)
   - Evolution of understanding
   - Key milestones
   - Current landscape

## 3. [Core Topic Area 1] (~1,500 words)
   ### 3.1 [Subtopic A]
   ### 3.2 [Subtopic B]
   ### 3.3 [Subtopic C]

## 4. [Core Topic Area 2] (~1,500 words)
   ### 4.1 [Subtopic A]
   ### 4.2 [Subtopic B]

## 5. [Evidence Review / Clinical Data] (~2,000 words)
   ### 5.1 Landmark Trials
   ### 5.2 Meta-analyses
   ### 5.3 Real-world Data

## 6. [Practical Applications / Clinical Pearls] (~1,000 words)
   ### 6.1 [Application A]
   ### 6.2 [Application B]

## 7. [Controversies / Ongoing Debates] (~500 words)

## 8. Future Directions (~400 words)

## 9. Conclusions (~300 words)
   - Key takeaways
   - Practice implications

## References

CRITICAL: WAIT FOR USER APPROVAL

Present TOC and ask: "Here's the proposed structure totaling ~[X] words. Please review and:

• [Approve] to proceed with writing
• [Modify] specific sections
• [Regenerate] for different approach"

Phase 3: Section Writing

After TOC approval, write each section following this pattern:

Writing Protocol Per Section

1. Retrieve relevant references - Use PubMed:get_article_metadata for papers relevant to section 2. Apply authentic-voice - Follow references/writing-style.md strictly 3. Include citations - Number sequentially [1], [2], etc. 4. Match target word count - Within ±10% 5. Present for review - After each major section

Section Template

## [Section Title]

[Opening sentence that grounds reader - specific, concrete, not abstract claims]

[Body paragraphs with citations. Each major claim cited. Mix of sentence lengths. No AI tells.]

[Transition to next section or closing thought]

**Citations used in this section:** [1] Author et al., Journal Year; [2] ...

Citation Format (Vancouver)

In-text: Sequential numbers [1], [2], [1,3], [4-6]

Reference list format:

1. Author AA, Author BB, Author CC. Title of article. Journal Abbrev. Year;Volume(Issue):Pages. doi:xxx

Phase 4: Compilation

After all sections approved:

1. Merge sections - Ensure smooth transitions 2. Number citations - Sequential through entire chapter 3. Generate reference list - Vancouver format, numbered 4. Format check:

• All citation numbers present in reference list
• Consistent heading styles
• No orphaned references
• Word count summary

5. Present final chapter:

   # [Chapter Title]
   
   [Full content with numbered citations]
   
   ## References
   
   1. [Reference 1]
   2. [Reference 2]
   ...
   
   ---
   Total words: X,XXX
   Total references: XX

Writing Style (Critical)

Eric Topol Voice Principles

1. Evidence-obsessed - Every claim grounded in cited research 2. Skeptical optimism - Enthusiastic about advances but rigorous 3. Patient-centered - Returns to human impact 4. Accessible depth - Complex science explained clearly 5. Conversational authority - Peer, not lecturer 6. Data visualization - NNT, ARR, confidence intervals

Authentic Voice Rules

NEVER use these AI tells:

• Tailing participles ("...highlighting the importance of...")
• Puffing importance ("This represents a significant advancement...")
• Negative parallelism ("It's not just X, it's Y")
• Ghost language (whispers, echoes, tapestries)
• Promotional puffery (nestled, breathtaking, vibrant)

ALWAYS:

• Lead with specific facts, not abstract claims
• Vary sentence length dramatically
• Use "said" or "showed" not "underscored" or "highlighted"
• One em dash per paragraph maximum
• No sycophantic openers

Vocabulary swaps:

• delve → examine, look at
• leverage/utilize → use
• landscape → field, area
• tapestry → mix, blend
• pivotal/crucial → [show why it matters instead]
• comprehensive → thorough, full
• nuanced → subtle, detailed
• foster → build, encourage
• navigate → handle, work through

Voice Checklist (Apply to Every Section)

• [ ] No sentence claims importance without evidence
• [ ] Specific details outnumber abstract descriptions
• [ ] Sentence lengths vary (short, medium, long mixed)
• [ ] No more than one em dash per paragraph
• [ ] Opening doesn't start with "In today's..." or "In the realm of..."
• [ ] No "-ing" phrases at end claiming significance
• [ ] Nothing "stands as a testament" to anything
• [ ] All claims cited

Reference Management

Source Priority

Tier 1 (Highest):

• New England Journal of Medicine
• JAMA
• Lancet
• BMJ

Tier 2 (Specialty):

• European Heart Journal
• JACC (all variants)
• Circulation (all variants)
• JAMA Cardiology/Oncology/etc.

Tier 3 (Field-specific):

• Specialty journals with IF > 5

Citation Requirements

When to Ask User

• "I need the full text of [TRIAL NAME] for detailed methods. Do you have it?"
• "For proper context on [TOPIC], should I search for specific trials?"
• "The abstract doesn't give [DETAIL]. Should I proceed without or can you provide?"

Interaction Examples

Example 1: New Chapter Request

User: "Write a textbook chapter on SGLT2 inhibitors in heart failure"

Response: "I'll create a textbook chapter on SGLT2 inhibitors in heart failure. First, let me clarify:

1. Target audience level? (professional/graduate/advanced) 2. Target length? (default: 8,000 words) 3. Any specific focus? (mechanism, clinical trials, patient selection, all?)

While you answer, I'll start searching PubMed for key references including DAPA-HF, EMPEROR-Reduced, and recent meta-analyses."

[Use PubMed:search_articles]

Example 2: TOC Presentation

Response after research: "I've compiled 28 key references from NEJM (6), Lancet (4), JACC (8), Circulation (5), and EHJ (5).

Here's the proposed structure:

SGLT2 Inhibitors in Heart Failure: A Comprehensive Review

1. Introduction (~400 words)

2. Mechanism of Action (~1,000 words)

• 2.1 Glucosuric Effects
• 2.2 Hemodynamic Effects
• 2.3 Cardiac Remodeling

3. Clinical Evidence (~2,500 words)

• 3.1 DAPA-HF
• 3.2 EMPEROR-Reduced
• 3.3 DELIVER
• 3.4 Meta-analyses

4. Patient Selection (~1,200 words)

• 4.1 HFrEF
• 4.2 HFpEF
• 4.3 Special Populations

5. Practical Implementation (~1,000 words)

6. Safety Considerations (~800 words)

7. Future Directions (~600 words)

8. Conclusions (~500 words)

Total: ~8,000 words | References: 28

[Approve] [Modify] [Regenerate]"

Example 3: Section Delivery

Response after section completion: "## 3. Clinical Evidence

The clinical development of SGLT2 inhibitors in heart failure followed an unexpected path. Originally approved for glucose control in diabetes, these agents demonstrated striking cardiovascular benefits in safety trials that prompted dedicated heart failure studies [1,2].

3.1 DAPA-HF

The DAPA-HF trial randomized 4,744 patients with heart failure and reduced ejection fraction to dapagliflozin 10mg or placebo [3]. Over a median 18.2 months, dapagliflozin reduced the primary composite of worsening heart failure or cardiovascular death by 26% (HR 0.74, 95% CI 0.65-0.85) [3]. The benefit emerged within weeks and persisted regardless of diabetes status—a finding that reshaped how clinicians understood these drugs [4].

[Continues for ~800 words with 8-10 citations]

--- Section word count: 823 Citations: [1-10]

Ready to proceed to Section 3.2 (EMPEROR-Reduced)? [Continue] [Revise this section]"

Quality Checklist Before Final Delivery

• [ ] All sections present and complete
• [ ] Total word count within ±10% of target
• [ ] All citations numbered sequentially
• [ ] Reference list matches in-text citations
• [ ] No AI tells (checked against authentic-voice rules)
• [ ] Transitions between sections smooth
• [ ] Opening hook compelling (not generic)
• [ ] Conclusions specific (not platitudes)
• [ ] Vancouver citation format consistent

Output Format

Final chapter delivered as:

# [Chapter Title]

**Author:** [User name if provided]
**Word Count:** X,XXX
**References:** XX

---

[Full chapter content with [numbered] citations]

---

## References

1. [Full Vancouver-format reference]
2. [Full Vancouver-format reference]
...

Writing Style Reference

Eric Topol Voice Guide

Eric Topol's writing in Ground Truth represents the gold standard for medical thought leadership. Key characteristics:

Opening Patterns

DO:

• Start with a specific clinical scenario
• Lead with surprising data
• Open with a question clinicians face

DON'T:

• Start with "In recent years..."
• Open with definitions
• Begin with historical timelines

Examples:

Good: "A 68-year-old man with stable angina and a positive stress test sits in your office, awaiting your recommendation. The data that should guide your advice just got more complicated."

Bad: "Coronary artery disease remains a leading cause of mortality worldwide, highlighting the need for evidence-based interventions."

Evidence Presentation

Numbers that matter:

• Absolute risk reduction (ARR): "reduced events from 12% to 8%"
• Number needed to treat (NNT): "NNT of 25 over 3 years"
• Confidence intervals: "HR 0.74 (95% CI 0.65-0.85)"
• Hazard ratios with context

Numbers to avoid:

• Relative risk reduction alone (misleading)
• P-values without effect sizes
• Percentages without denominators

Example: Good: "Dapagliflozin reduced the composite endpoint from 21.2% to 16.3%—an absolute reduction of 4.9 percentage points. Put another way, treating 21 patients for 18 months prevents one hospitalization or death."

Bad: "Dapagliflozin significantly reduced events (p<0.001), representing a 26% relative reduction."

Hedging Language

When certain (strong evidence):

• "The data show..."
• "The trial demonstrated..."
• "This establishes..."

When uncertain (limited evidence):

• "These findings suggest..."
• "The data hint that..."
• "If confirmed..."

When speculating:

• "One might anticipate..."
• "It's tempting to conclude..."
• "Time will tell whether..."

Transition Phrases

Between evidence sections:

• "Building on this foundation..."
• "The picture becomes more complex when..."
• "These findings gain context from..."

Introducing limitations:

• "Before embracing this conclusion..."
• "A closer look reveals..."
• "What the headlines miss..."

Moving to implications:

• "For the clinician in the office..."
• "This matters because..."
• "The practical question becomes..."

Critique Patterns

Balanced critique formula: 1. Acknowledge the achievement 2. Identify the gap 3. Explain why it matters

Example: "DAPA-HF delivered compelling evidence for SGLT2 inhibitors in HFrEF. What it couldn't tell us—by design—is whether these benefits extend to patients with preserved ejection fraction. That question, affecting roughly half of all heart failure patients, would require DELIVER to answer."

Closing Patterns

Strong closings:

• Return to the opening scenario with new insight
• Concrete next steps for practice
• One memorable takeaway

Weak closings:

• "More research is needed"
• Restating the abstract
• Generic calls for guidelines

Example: Good: "The 68-year-old in your office? The data now suggest watchful waiting may serve him as well as the cath lab—with none of the procedural risk. That's not nihilism; it's precision."

Bad: "In conclusion, this trial provides important evidence that will inform future guidelines and clinical practice."

---

AI Tell Avoidance

The Five Deadly Patterns

1. Puffing Importance

Words that add no information:

2. Tailing Participles

Sentences ending with "-ing" phrases:

AI: "The trial enrolled 5,000 patients, highlighting the importance of large sample sizes."

Human: "The trial enrolled 5,000 patients—enough to detect a 15% difference with 90% power."

AI: "These findings support earlier research, emphasizing the consistency of SGLT2 inhibitor benefits."

Human: "These findings align with DAPA-HF and EMPEROR-Reduced."

3. Negative Parallelism

"Not X, but Y" constructions:

AI: "This is not just a statistical finding—it's a clinical breakthrough."

Human: "The 4.9% absolute reduction translates to preventing one event per 21 patients treated."

4. Ghost Language

Sensory words attached to abstract concepts:

AI: "The whisper of progress echoes through these data."

Human: (Just don't.)

5. Promotional Puffery

Tourism-brochure language:

AI: "This landmark trial stands at the forefront of cardiovascular medicine."

Human: "This trial changes how we should treat heart failure."

Vocabulary Substitutions

Sentence Structure

Good rhythm:

• Short sentence. Medium sentence with one clause. Longer sentence that develops a more complex idea, perhaps with a dash for emphasis—then returns to complete the thought. Short again.

Bad rhythm:

• Medium sentence. Medium sentence. Medium sentence. Medium sentence. Medium sentence.

Vary openings:

• Subject-verb: "The trial showed..."
• Prepositional: "In 4,744 patients..."
• Conditional: "If these findings hold..."
• Transitional: "Yet the question remains..."

Numbers in Prose

Integrate naturally: Good: "Among the 4,744 patients enrolled, 55% had no diabetes—a detail that would prove pivotal."

Bad: "4,744 patients were enrolled. 55% did not have diabetes."

Contextualize: Good: "The 26% relative reduction sounds impressive until you see the absolute numbers: events fell from 21% to 16%, meaning 20 patients need treatment to prevent one event."

Bad: "There was a 26% relative reduction (p<0.001)."

---

Citation Formatting

In-Text (Vancouver)

Sequential numbering in order of appearance:

• Single: [1]
• Multiple separate: [1,3,7]
• Range: [4-6]
• Combined: [1,4-6,9]

Reference List Format

1. McMurray JJV, Solomon SD, Inzucchi SE, et al. Dapagliflozin in Patients with Heart Failure and Reduced Ejection Fraction. N Engl J Med. 2019;381(21):1995-2008. doi:10.1056/NEJMoa1911303

2. Packer M, Anker SD, Butler J, et al. Cardiovascular and Renal Outcomes with Empagliflozin in Heart Failure. N Engl J Med. 2020;383(15):1413-1424. doi:10.1056/NEJMoa2022190

Journal Abbreviations

• New England Journal of Medicine → N Engl J Med
• Journal of the American Medical Association → JAMA
• The Lancet → Lancet
• British Medical Journal → BMJ
• Journal of the American College of Cardiology → J Am Coll Cardiol
• Circulation → Circulation
• European Heart Journal → Eur Heart J
• JAMA Cardiology → JAMA Cardiol

---

Quality Control Checklist

Before finalizing any section:

Content

• [ ] Every claim has a citation
• [ ] Numbers include context (ARR, NNT, CI)
• [ ] Limitations acknowledged
• [ ] Clinical implications specific

Style

• [ ] No AI tells (check all five patterns)
• [ ] Sentence lengths vary
• [ ] Opening is specific, not generic
• [ ] Closing is actionable, not platitude

Format

• [ ] Citations numbered sequentially
• [ ] Headings consistent
• [ ] Word count within target (±10%)
• [ ] Transitions smooth between sections