PubMed Search outputs

The search returns matched article metadata — PMIDs, titles, abstracts, authors, and journal details. Those abstracts and PMIDs feed model-driven entity extraction, a Literature Passport, or Semantic Scholar citation analysis.

Output structure

PubMed Search returns a JSON object containing matched article metadata and search summary information.

Top-level fields

Article fields

Each article in the articles array contains:

Plain-search additional fields

A plain search calls EFetch alongside ESummary, so its results carry more than the basic fields above when present:

Filtered-search fields

A search filtered by author, affiliation, date range, or article type only calls ESearch + ESummary — no EFetch — so its results have no abstract, mesh_terms, or keywords. It returns pmid, title, authors, journal, year, doi, pmcid, and a has_fulltext boolean (true when a pmcid is present).

Example output

{
  "total": 1247,
  "query": "EGFR resistance osimertinib",
  "articles": [
    {
      "pmid": "38456789",
      "title": "Mechanisms of acquired resistance to osimertinib in EGFR-mutant NSCLC",
      "abstract": "Acquired resistance to osimertinib remains a major clinical challenge...",
      "authors": ["Wang S", "Chen Y", "Li X", "Zhang H", "Liu M"],
      "journal": "J Clin Oncol",
      "year": "2025",
      "doi": "10.1200/JCO.2025.00000",
      "mesh_terms": ["Protein Kinase Inhibitors", "Carcinoma, Non-Small-Cell Lung", "Drug Resistance, Neoplasm"],
      "pubmed_url": "https://pubmed.ncbi.nlm.nih.gov/38456789/"
    },
    {
      "pmid": "38234567",
      "title": "C797S mutation and combination strategies for EGFR TKI resistance",
      "abstract": "",
      "authors": ["Kim J", "Park S", "Lee D"],
      "journal": "Cancer Res",
      "year": "2025",
      "pubmed_url": "https://pubmed.ncbi.nlm.nih.gov/38234567/"
    }
  ]
}

The second article shows an empty abstract — EFetch returns no <AbstractText> for every PMID (older records and some non-English-language articles commonly lack one), and the search does not fabricate one.

Canvas visualization

PubMed Search results render through a custom canvas renderer as a list of article titles with journal and year, following the platform's standard card shape — there's no PubMed-specific card size or border color.

Interpreting canvas metrics

Downstream data use

Entity extraction

Ask the model to read the returned abstracts directly — no separate tool call. It extracts:

• Gene/protein names: EGFR, TP53, BRCA1
• Compound names: osimertinib, gefitinib, sotorasib
• Disease terms: NSCLC, breast cancer, AML
• Relationships: Drug-target, drug-disease, gene-disease associations

Chaining into Literature Passport

The Literature Passport aggregates PubMed results with entity extraction and trend analysis into a comprehensive report. It uses the PubMed result set as its primary data source.

Chaining into Semantic Scholar

Pass PMIDs to Semantic Scholar for citation network analysis, influential citation identification, and citation count data that PubMed does not provide natively.

Error responses

When a search fails or returns no results:

Common error values:

• "No results found" — Query matched zero articles
• "NCBI API timeout" — E-utilities did not respond within 45 seconds
• "Rate limit exceeded" — Too many concurrent NCBI requests

Read next