Back to Details

synthesis-writer

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Synthesis Writer (systematic review)

Synthesis Writer（系统综述）

Goal: write a structured synthesis that is traceable back to extracted data.
目标：撰写可追溯至提取数据的结构化证据合成内容。

Role cards (use explicitly)

角色卡片（需明确使用）

Evidence Synthesizer (table-driven)

证据合成师（表格驱动）

Mission: turn extracted rows into comparative findings without inventing claims.
Do:
  • Summarize the included evidence base with counts and basic descriptors from the table.
  • Group studies by theme/intervention/outcome using extraction fields (not impressions).
  • Report agreements/disagreements and heterogeneity explicitly.
Avoid:
  • Conclusions that are not supported by fields present in the table.
  • Overconfident language when bias/heterogeneity is high.
任务：将提取的表格行内容转化为对比性研究结果，不得凭空生成结论。
需执行：
  • 利用表格中的统计数量与基础描述符，总结纳入的证据库。
  • 依据提取字段（而非主观印象）按主题/干预措施/结果对研究进行分组。
  • 明确报告研究间的一致性/分歧点与异质性。
需避免：
  • 得出表格提取字段未支撑的结论。
  • 当偏倚/异质性较高时使用过于绝对的表述。

Bias Reporter (skeptic)

偏倚报告员（持怀疑态度）

Mission: keep conclusions bounded by risk-of-bias and missing data.
Do:
  • Summarize RoB patterns and how they affect interpretation.
  • Separate "supported" vs "needs more evidence" statements.
Avoid:
  • Generic boilerplate; tie limitations to observed gaps (missing baselines, protocol differences, etc.).
任务：确保结论受限于偏倚风险与数据缺失情况。
需执行：
  • 总结偏倚风险（RoB）的模式及其对结果解读的影响。
  • 区分“有证据支撑”与“需更多证据验证”的表述。
需避免：
  • 使用通用套话；需将局限性与观察到的空白（如缺失基线数据、研究方案差异等）关联。

Role prompt: Systematic Review Synthesizer

角色提示词：系统综述合成师

text
You are writing the synthesis section of a systematic review.

Your job is to produce a narrative that is traceable back to papers/extraction_table.csv:
- describe the evidence base
- synthesize findings by theme
- report heterogeneity and disagreements
- state limitations and risk-of-bias implications

Constraints:
- do not invent facts beyond the extraction table
- if a claim cannot be backed by extracted fields, mark it as a verification need or remove it

Style:
- structured, comparative, cautious
text
You are writing the synthesis section of a systematic review.

Your job is to produce a narrative that is traceable back to papers/extraction_table.csv:
- describe the evidence base
- synthesize findings by theme
- report heterogeneity and disagreements
- state limitations and risk-of-bias implications

Constraints:
- do not invent facts beyond the extraction table
- if a claim cannot be backed by extracted fields, mark it as a verification need or remove it

Style:
- structured, comparative, cautious

Inputs

输入项

Required:
  • papers/extraction_table.csv
Optional:
  • DECISIONS.md
    (approval to write prose, if your process requires it)
  • output/PROTOCOL.md
    (to restate scope and methods consistently)
必填：
  • papers/extraction_table.csv
可选：
  • DECISIONS.md
    （若流程要求，需此文档确认可开始撰写）
  • output/PROTOCOL.md
    （用于保持研究范围与方法的一致性）

Outputs

输出项

  • output/SYNTHESIS.md
  • output/SYNTHESIS.md

Workflow

工作流程

  1. Check writing approval (if applicable)
    • If your pipeline requires it, confirm 
      DECISIONS.md
      indicates approval before writing prose.
  2. Describe the evidence base (methods snapshot)
    • Summarize the included set using 
      papers/extraction_table.csv
      (counts, time window, study types).
    • Keep this strictly descriptive.
  3. Theme-based synthesis
    • Group studies by theme/intervention/outcome (based on extraction fields).
    • For each theme, compare results across studies and highlight disagreements/heterogeneity.
  4. Bias + limitations
    • Summarize RoB patterns using the bias fields in 
      papers/extraction_table.csv
      .
    • Call out limitations that block strong conclusions (missing baselines, weak measures, publication bias signals).
  5. Conclusions (bounded)
    • State only what the extracted evidence supports.
    • Separate “supported conclusions” vs “needs more evidence”.
  1. 检查撰写许可（如适用）
    • 若流程要求，需确认
      DECISIONS.md
      中已批准撰写后再开始。
  2. 描述证据库（方法概要）
    • 利用
      papers/extraction_table.csv
      总结纳入的研究集合（数量、时间范围、研究类型）。
    • 仅做客观描述。
  3. 基于主题的证据合成
    • 依据提取字段按主题/干预措施/结果对研究进行分组。
    • 针对每个主题，对比不同研究的结果，突出分歧点与异质性。
  4. 偏倚与局限性
    • 利用
      papers/extraction_table.csv
      中的偏倚字段总结RoB模式。
    • 指出会影响结论可靠性的局限性（如缺失基线数据、测量方法薄弱、发表偏倚信号等）。
  5. 结论（受限表述）
    • 仅阐述提取证据所支撑的内容。
    • 区分“有证据支撑的结论”与“需更多证据验证的内容”。

Mini examples (traceability)

示例（可追溯性）

  • Bad (untraceable): 
    Most studies show large improvements.
  • Better (table-driven): 
    Across the included studies (n=...), reported success rates improve in ... settings; however, protocols vary (tool access, budgets), and several studies omit ... fields, limiting comparability.
  • Bad (generic limitation): 
    There may be publication bias.
  • Better (specific): 
    Few studies report negative results or failed runs; combined with sparse ablation reporting, this raises the risk that improvements are protocol- or tuning-dependent.
  • 反面示例（不可追溯）：
    大多数研究显示显著改善。
  • 正面示例（基于表格）：
    在纳入的研究中（n=...），报告的成功率在...场景下有所提升；但研究方案存在差异（工具可及性、预算等），且部分研究未提供...字段，限制了可比性。
  • 反面示例（通用局限性）：
    可能存在发表偏倚。
  • 正面示例（具体表述）：
    极少研究报告阴性结果或失败案例；结合稀疏的消融实验报告，这提升了研究效果依赖于方案或调优的风险。

Suggested outline for 
output/SYNTHESIS.md

output/SYNTHESIS.md
建议大纲

  • Research questions + scope (from 
    output/PROTOCOL.md
    )
  • Methods (sources, screening, extraction)
  • Included studies summary (table-driven)
  • Findings by theme (table-driven)
  • Risk of bias + limitations
  • Implications + future work (bounded)
  • 研究问题与范围（来自
    output/PROTOCOL.md
    ）
  • 研究方法（数据源、筛选、提取步骤）
  • 纳入研究总结（基于表格）
  • 按主题分类的研究结果（基于表格）
  • 偏倚风险与局限性
  • 研究启示与未来工作（受限表述）

Definition of Done

完成标准

  • Every major claim in 
    output/SYNTHESIS.md
    is traceable to specific fields/rows in 
    papers/extraction_table.csv
    .
  • Limitations and bias considerations are explicit (not generic boilerplate).
  • output/SYNTHESIS.md
    中的每个主要结论均可追溯至
    papers/extraction_table.csv
    中的特定字段/行。
  • 局限性与偏倚考量明确具体（非通用套话）。

Troubleshooting

问题排查

Issue: the synthesis starts inventing facts not in the table

问题：合成内容开始生成表格中未提及的信息

Fix:
  • Restrict claims to what is explicitly present in 
    papers/extraction_table.csv
    ; move speculation to “needs more evidence”.
解决方法：
  • 仅基于
    papers/extraction_table.csv
    中明确存在的内容提出结论；将推测内容移至“需更多证据验证”部分或删除。

Issue: extraction table is too sparse to synthesize

问题：提取表格内容过于稀疏，无法进行合成

Fix:
  • Add missing extraction fields/values first (re-run 
    extraction-form
    / 
    bias-assessor
    ), then write.
解决方法：
  • 先补充缺失的提取字段/值（重新运行
    extraction-form
    / 
    bias-assessor
    ），再进行撰写。