NNLM webinar walks users through PubMed search tools; NLM staff answer common questions

Margie Sheppard of the Network of the National Library of Medicine opened a webinar demonstrating how PubMed organizes and returns biomedical literature, walking attendees through citation searches, author‑search techniques, filters and saving tools. Amanda Sawyer, a representative from the National Library of Medicine (NLM), joined to explain how content is selected for PubMed and to field audience questions.

Why it matters: PubMed is a primary access point for clinicians, researchers and students seeking biomedical citations. Understanding how PubMed maps keywords to Medical Subject Headings (MeSH), identifies authors and links to full text helps users locate relevant literature more efficiently.

Sheppard emphasized that PubMed is “a database of citations, not a database of full‑text articles,” and that while roughly 75% of citations include links to full text, access depends on publisher permissions and whether content is deposited in PubMed Central (PMC). Sawyer summarized the main pathways for content to appear in PubMed: journal selection into MEDLINE after scientific and technical review, participation in PubMed Central where journals deposit free full text, and author manuscripts deposited under the NIH public access policy for NIH‑funded research. “The decision is typically made at the level of the journal,” Sawyer said when describing selection into MEDLINE.

The session included live demonstrations with concrete examples: a citation search by journal/year/author that triggered PubMed’s citation sensor and an abstract view showing the PM ID (PubMed identifier) and DOI links; an author search workflow using the recommended “last name + first initial” format and PubMed’s computed author sort for author disambiguation; and keyword searches that illustrated automatic term mapping (ATM), where user terms such as “nosebleed” map to the MeSH term “epistaxis.” Sheppard noted that ATM adds MeSH terms when available so users need not know technical vocabulary to retrieve comprehensive results.

Sheppard also reviewed result‑management features. Registered users with a MyNCBI account can create saved searches and alerts, build collections, email citation lists, export .nbib files for citation managers, and generate RSS feeds. The presenters demonstrated filters that narrow large result sets by language, publication date, free full text, article type (for example, clinical trial or systematic review), and an associated‑data filter that surfaces citations linking to secondary data repositories such as GenBank, ClinicalTrials.gov, Figshare or Dryad.

On sorting, Sheppard explained that the default “best match” ranking is a machine‑learning algorithm designed to surface the most relevant citations first, while “most recent” orders results by when they were added to PubMed. She advised users that changing sort order does not change the underlying result set, only its ordering.

During Q&A, presenters answered audience questions about historical coverage (PubMed records can date back to the 18th century for digitized historical journals), the difference between MEDLINE and PMC (only MEDLINE is indexed with MeSH), the linkage between ClinicalTrials.gov records and PubMed citations, and whether PubMed offers official AI integrations (presenters said none are officially offered at this time). When asked whether PubMed includes full text by default, the session reiterated the distinction between citations and full text and pointed participants to PMC when free full text is available.

The webinar closed with logistics for the recorded session, a link to the evaluation and a reminder of two follow‑up sessions on MeSH and automatic term mapping led by Margot Malachowski. Presenters invited attendees to use the NLM support center or submit help tickets for complex or site‑specific issues.

The session did not include formal decisions or policy votes; it was an educational demonstration and Q&A.