為什麼這三者決定能見度？
大部分讀者：搜尋 → 看標題 → 點進去 → 讀摘要 → 決定要不要下載全文。能讓你的論文被檢索到、被決定下載的，就只有標題、摘要、關鍵字三項——它們的索引被 PubMed、Google Scholar、bioRxiv 收錄，全文反而不在第一波檢索結果中。
所以：標題與摘要應在投稿前最後再寫一次，把已知的最強亮點放上去。
Most readers: search → see title → click → read abstract → decide whether to download. Only title, abstract, and keywords are searchable in PubMed, Google Scholar, bioRxiv; the full text isn't returned in the first wave of search.
So: rewrite title and abstract one last time before submission, packing in your strongest hooks.
一、好標題的 5 個公式
| 公式 | 範例 |
|---|---|
| ① 結果型 | "Single-cell profiling identifies three chemoresistance-associated cell states in triple-negative breast cancer""Single-cell profiling identifies three chemoresistance-associated cell states in triple-negative breast cancer" |
| ② 方法型 | "scIBD: a benchmarking framework for single-cell integration with batch effects" |
| ③ 比較型 | "Pembrolizumab vs chemotherapy as first-line treatment for advanced NSCLC: a randomized phase III trial""Pembrolizumab vs chemotherapy as first-line treatment for advanced NSCLC: a randomized phase III trial" |
| ④ 機制型 | "IFN-γ induces ISG-driven epigenetic reprogramming via STAT1-mediated chromatin remodeling" |
| ⑤ 問題型 | "Does post-chemotherapy minimal residual disease predict relapse in TNBC? A prospective multi-center cohort" |
二、結構式 vs 非結構式
📋 結構式摘要
分小標：Background / Methods / Results / Conclusions。常見限制 250-350 字。
適用：NEJM, Lancet, JAMA, BMJ, BMC 系列、PLOS Medicine、大部分臨床期刊。
With sub-headings: Background / Methods / Results / Conclusions. Usually 250-350 words.
Used by: NEJM, Lancet, JAMA, BMJ, BMC family, PLOS Medicine, most clinical journals.
📜 非結構式摘要
單段連續敘事。Cell, Nature, Science 系列偏好；通常 150-200 字。
適用：Nature 系列、Cell, Science、PNAS、許多 basic-science 期刊。
Single continuous paragraph. Preferred by Nature, Cell, Science families; usually 150-200 words.
Used by: Nature family, Cell, Science, PNAS, many basic-science journals.
三、「1+2+3+1」摘要公式
把 250 字平均分成四塊：
Distribute 250 words across four blocks:
Background (1 句)
領域問題 + 既有缺口（不要超過 1 句）。例：「TNBC 缺乏標靶治療，且化療抗性的細胞層級機制仍未明。」
Field problem + gap (one sentence). Eg: "TNBC lacks targeted therapy and the cell-level mechanism of chemoresistance remains unclear."
Methods (2 句)
研究設計 + 樣本量 + 關鍵方法。例：「我們對 18 例 TNBC 治療前後配對組織進行 scRNA-seq (10x v3)。透過 trajectory analysis 鑑定化療誘導的細胞態轉變。」
Design + n + key methods. Eg: "We performed scRNA-seq (10x v3) on 18 paired pre/post-chemotherapy TNBC samples. Trajectory analysis identified chemo-induced cell-state transitions."
Results (3 句)
主要發現 + 數字 + 至少一個效應量。例：「我們鑑定 3 個化療抗性相關細胞態 (狀態 R1-R3)。R2 在化療後上升 12 倍 (p<0.001) 並與 6 個月內復發顯著相關 (HR 2.4, 95% CI 1.5-3.8)。R2 高表達 ALDH1A3 與 EMT 標誌基因。」
Main findings + numbers + ≥1 effect size. Eg: "We identified 3 chemoresistance-associated cell states (R1-R3). R2 rose 12-fold post-chemo (p<0.001) and was associated with 6-month relapse (HR 2.4, 95% CI 1.5-3.8). R2 highly expressed ALDH1A3 and EMT markers."
Conclusion (1 句)
對領域 / 臨床的意義 (校準語氣)。例：「ALDH1A3+ EMT-high 細胞態可能是 TNBC 化療抗性的新標靶，值得後續功能驗證。」
Implication (calibrated). Eg: "ALDH1A3+ EMT-high cell states may represent a novel chemoresistance target in TNBC and warrant functional validation."
四、摘要對照
❌ 模糊版
「乳癌是常見癌症。我們做了單細胞分析。我們找到一些有趣的結果。這些結果可能對未來研究有幫助。」
"Breast cancer is a common cancer. We did single-cell analysis. We found some interesting results. These may help future research."
✅ 具體版
「Background: 三陰性乳癌 (TNBC) 缺乏標靶治療，化療抗性細胞層級機制不明。Methods: 我們對 18 例配對 pre/post 化療 TNBC 進行 scRNA-seq (10x v3)，以 trajectory analysis 鑑定化療誘導之細胞態轉變。Results: 鑑定出 3 個化療抗性細胞態 (R1-R3)；R2 在化療後上升 12 倍 (p<0.001)，與 6 個月內復發顯著相關 (HR 2.4, 95% CI 1.5-3.8)；R2 表達 ALDH1A3 與 EMT 基因。Conclusion: ALDH1A3+ EMT-high 細胞態可能為 TNBC 化療抗性新標靶，值得功能驗證。」
"Background: TNBC lacks targeted therapy and the cell-level basis of chemoresistance is unclear. Methods: We performed scRNA-seq (10x v3) on 18 paired pre/post-chemo TNBC samples; trajectory analysis identified chemo-induced cell-state transitions. Results: 3 chemoresistance-associated states (R1-R3); R2 rose 12-fold post-chemo (p<0.001) and was associated with 6-month relapse (HR 2.4, 95% CI 1.5-3.8); R2 expressed ALDH1A3 and EMT markers. Conclusion: ALDH1A3+ EMT-high states may represent a novel chemoresistance target warranting functional validation."
五、關鍵字 SEO 與 MeSH
大多期刊要求 4-7 個關鍵字。策略：
- 優先用 MeSH terms（在 NCBI MeSH Browser 查），確保 PubMed 索引到位。
- 不要重複標題裡已有的字（系統已索引），補上「同義詞 / 上位詞 / 應用領域」。
- 包含 1-2 個「方法學」關鍵字（如 single-cell RNA sequencing），讓做方法搜尋的讀者找到。
- 包含 1 個「應用領域」關鍵字（如 chemotherapy resistance, biomarker discovery）。
Most journals ask for 4-7 keywords. Strategy:
- Prefer MeSH terms (lookup at NCBI MeSH Browser) to ensure PubMed indexing.
- Don't duplicate words already in the title (already indexed); add synonyms / parent terms / application areas.
- Include 1-2 method-level keywords (e.g., single-cell RNA sequencing) to be found by methodologists.
- Include 1 application area (e.g., chemotherapy resistance, biomarker discovery).
| 範例文章 | 標題已有 | 補充關鍵字 |
|---|---|---|
| scRNA-seq + TNBC + chemoresistancescRNA-seq + TNBC + chemoresistance | single-cell, TNBC, chemoresistance | cancer stem cells, ALDH1A3, EMT, tumor heterogeneity, breast neoplasms (MeSH), drug resistance neoplasm (MeSH) |
六、最後 30 分鐘檢查
🌳 送審前自檢
📝 自我檢測
1. 哪一個是「結果型」標題的最佳範例？
1. Which is the best "findings-type" title?
2. 結構式摘要的主要使用場域是？
2. Structured abstracts are primarily used by?
3. 關於關鍵字選擇，下列何者最正確？
3. Best practice for keyword selection?