輸入檢索詞 → 從候選清單選擇 MeSH 控制詞彙 → 一鍵加入同義詞 (Entry Terms) → 組成 PubMed PICO+研究設計之檢索策略Enter one term → pick the matching MeSH heading → tick Entry Terms as free-text → build a PubMed PICO + study-design strategy

① PICO + 研究設計① PICO + Study design

在任一列輸入英文檢索詞並從候選清單選擇 MeSH 標目,會同時加入控制辭彙與其同義詞(Entry Terms);點標目標籤可在右側檢視上下位結構,點 ▾ 可切換欄位。拖曳左側 ⠿ 可調整列的順序。同一列內以 OR 連結,右上角有勾選的列(加入檢索策略)之間以 AND 組合。PICO 只是引導框架:列的代號(如 P1)與名稱可直接點選修改,也可新增或刪除概念列。Type an English term in any row and pick a MeSH heading — the heading and its synonyms (Entry Terms) are added together; click a heading tag to see its hierarchy on the right, or ▾ to change its field. Drag ⠿ to reorder rows. Terms within a row are ORed; rows ticked at the top right (added to the strategy) are ANDed. PICO is only a guide: click a row's code (e.g. P1) or name to rename it, and add or delete rows as needed.

② PubMed 檢索策略② PubMed search strategy

在 PubMed 執行 ↗Run in PubMed ↗

使用提示Tips

  • 控制辭彙(含狹義詞)[mh] 會自動包含樹狀結構中所有狹義(下位)標目;不含狹義詞 [mh:noexp] 則只找此標目。重點控制辭彙 [majr] 或 [majr:noexp] 只找以此為文章重點的文獻,並分別含或不含所有下位詞。
  • Entry Terms 常是倒裝形式(例:Attack, Heart),加入時會自動轉成自然語序 Heart Attack,才能在題名/摘要中比對到。
  • 找不到標目時(例:縮寫 GDM),會自動取 PubMed 相關性前 50 篇文章,統計它們最常標的 MeSH,列出前 10 名並標示出現篇數;名稱有對到但不是想要的,可按候選清單底部的「反推」。
  • 補充概念 [nm]:藥物、化學物質、罕見疾病等多為 MeSH 的補充概念,不在標目樹中,標籤顯示「補充概念」並以 "xxx"[nm] 檢索;右側若列出對應標目,建議一併加入。
  • 鍵盤刪除:輸入框空白時,用 ← → 在詞彙間移動(紅框標示),按 Backspace 或 Delete 刪除紅框詞彙,可連續刪除;也可直接點詞彙選取。只按 Backspace 時,第一下會先標示最後一個詞彙。刪除標目時連同其同義詞一起移除。
  • 更多欄位選項:點「更多欄位選項」可新增以 PubMed 欄位檢索的列(作者 [au]、期刊 [ta]、PMID [pmid]、機構 [ad] 等,依屬性分組);此列輸入的詞一律用該欄位檢索,不查 MeSH,例:Smith J[au]。
  • 鄰近字:每列可設定一次,套用到該列所有兩個字以上的自由詞彙,例:"hip fracture"[tiab:~3] 會找到 fracture of the hip。PubMed 鄰近字只支援篇名/摘要 [tiab] 與篇名 [ti];選所有欄位 [all] 或文字欄位 [tw] 時不套用,且不能與截字 * 併用。
  • 出版類型 [pt]:研究設計欄位加入 Randomized Controlled Trial、Meta-Analysis 等標目時預設為 [pt](例:"Randomized Controlled Trial"[pt])。請同時放自由詞彙(randomized、placebo),以免漏掉尚未標引的新文獻。
  • 比較(C)若與介入相同概念,通常不必納入 AND 組合,可取消該列右上角的勾選。
  • MeSH (incl. narrower) [mh] explodes to every narrower heading in the tree; no narrower [mh:noexp] searches this heading only. Major MeSH [majr] or [majr:noexp] finds only articles where it is a main focus, with or without all narrower headings respectively.
  • Entry Terms are often inverted (e.g. Attack, Heart); they are automatically converted to natural order (Heart Attack) so they match titles/abstracts.
  • No heading found (e.g. the abbreviation GDM): the tool takes the top 50 relevance-ranked PubMed articles, tallies the MeSH headings they carry and lists the top 10 with article counts. If a name match is not what you meant, use the “infer” button at the bottom of the suggestion list.
  • Supplementary concepts [nm]: drugs, chemicals and rare diseases are often MeSH supplementary concepts outside the heading tree. They show a “supplementary” badge and are searched as "xxx"[nm]; if mapped headings are listed on the right, consider adding them too.
  • Keyboard deletion: with the input box empty, use ← → to move between terms (red outline) and press Backspace or Delete to remove the outlined one; keep pressing to remove several in a row. You can also click a term to select it. Pressing Backspace alone first outlines the last term. A heading is removed together with its synonyms.
  • More field options: add a row that searches a specific PubMed field (Author [au], Journal [ta], PMID [pmid], Affiliation [ad]…, grouped by kind). Every term in that row is searched in that field and no MeSH lookup is made, e.g. Smith J[au].
  • Proximity: set once per row; it applies to every multi-word free-text term in that row, e.g. "hip fracture"[tiab:~3] also finds fracture of the hip. PubMed proximity works only in [tiab] and [ti] (not applied with [all] / [tw]), and not with truncation *.
  • Publication type [pt]: in Study design, headings such as Randomized Controlled Trial or Meta-Analysis default to [pt] (e.g. "Randomized Controlled Trial"[pt]). Add free-text too (randomized, placebo) to catch not-yet-indexed records.
  • If the Comparison duplicates the Intervention concept, untick the box at the top right of that row so it is not ANDed.