PubMed Scraper

PubMed Scraper extracts biomedical articles from pubmed.ncbi.nlm.nih.gov through NCBI's public E-utilities API: search by keyword (the full PubMed query syntax works) or look up PMIDs, and get the title, abstract, authors with affiliations, journal, dates, DOI, PMCID, publication types, MeSH terms and keywords as one flat JSON record per article. No browser, no login, no API key required.

What you get

Field	Example
pmid, title, url	36319653, "Resting-state BOLD signal variability…", https://pubmed.ncbi.nlm.nih.gov/36319653/
abstract, hasAbstract	Full abstract; structured abstracts keep their labels (BACKGROUND: … METHODS: …)
authors[], authorNames[], firstAuthor, lastAuthor, authorCount	Names as printed in the article, initials, ORCID when PubMed has it, per-author affiliations
affiliations[]	Distinct affiliation strings (e-mail addresses removed)
journal, journalAbbrev, issn, eissn, journalCountry, nlmJournalId	"Scientific reports", "Sci Rep", 2045-2322, England
volume, issue, pages, citation	12, 1, 18425, "Sci Rep. 2022 Nov 01;12(1):18425. doi: 10.1038/…"
publicationDate, publicationYear, epubDate, pubmedDate	ISO dates, as precise as PubMed has them (2022-11-01, 2017-06, 2019)
doi, doiUrl, pmcid, pmcUrl, pii	10.1038/s41598-022-21703-5, PMC9626555 (free full text)
publicationTypes[], isReview, language, publicationStatus, medlineStatus	["Journal Article", "Review"], true, eng, epublish, MEDLINE
meshTerms[], meshMajorTopics[], keywords[], chemicals[], fundingAgencies[], referenceCount	MeSH indexing, author keywords, substances, grant agencies
source, searchTerm, position	Which search produced the record and its rank (Best Match or newest first)

NCBI Bookshelf chapters indexed in PubMed (GeneReviews, StatPearls…) come back with the same keys plus bookTitle, bookAccession and publisher (recordType: "book").

Use cases

• Literature monitoring: run a saved query weekly with dateFrom and sort: pub_date to catch new papers on a drug, gene or disease.
• Systematic reviews and meta-analyses: pull thousands of records with abstracts and MeSH terms into a spreadsheet or a screening tool.
• RAG and AI agents: feed abstracts, keywords and DOIs to an LLM pipeline; every record carries a stable PMID and DOI for citations.
• Bibliometrics: journals, countries, publication types, funding agencies and author counts per topic and year.
• Enrichment: give it a list of PMIDs from a reference manager and get back clean metadata, PMCIDs and DOIs.

Input

{
    "searchTerms": ["crispr gene editing", "\"Diabetes Mellitus, Type 2\"[mesh] AND review[pt]"],
    "pmids": ["36319653", "https://pubmed.ncbi.nlm.nih.gov/31452104/"],
    "maxItems": 200,
    "sort": "pub_date",
    "dateFrom": "2024-01-01",
    "dateTo": "2024-12-31",
    "includeAbstract": true
}

• searchTerms accept anything the PubMed search box accepts: boolean operators, field tags ([ti], [au], [mesh], [pt], [sb]), phrases in quotes. PubMed pages at most 10,000 results per query; split big queries by date.
• pmids are fetched first, 200 per request, in the order given.
• maxItems is exact and global; an article found by several terms is saved and charged once.
• sort: relevance (PubMed Best Match, default) or pub_date (newest first).
• dateFrom / dateTo filter by publication date (YYYY-MM-DD, YYYY-MM or YYYY); either side may be left empty.
• apiKey (optional): your free NCBI key raises NCBI's limit from 3 to 10 requests per second. The key is only sent to eutils.ncbi.nlm.nih.gov.

Output example

{
    "pmid": 31452104,
    "title": "Molegro Virtual Docker for Docking.",
    "abstract": "Molegro Virtual Docker is a protein-ligand docking simulation program that allows us to carry out docking simulations in a fully integrated computational package. …",
    "hasAbstract": true,
    "authors": [
        { "fullName": "Gabriela Bitencourt-Ferreira", "lastName": "Bitencourt-Ferreira", "foreName": "Gabriela", "initials": "G", "collectiveName": null, "orcid": null, "affiliations": ["Escola de Ciências da Saúde, Pontifícia Universidade Católica do Rio Grande do Sul-PUCRS, Porto Alegre, RS, Brazil."] },
        { "fullName": "Walter Filgueira de Azevedo", "lastName": "de Azevedo", "foreName": "Walter Filgueira", "initials": "WF", "collectiveName": null, "orcid": null, "affiliations": ["Escola de Ciências da Saúde, Pontifícia Universidade Católica do Rio Grande do Sul-PUCRS, Porto Alegre, RS, Brazil."] }
    ],
    "authorNames": ["Gabriela Bitencourt-Ferreira", "Walter Filgueira de Azevedo"],
    "firstAuthor": "Gabriela Bitencourt-Ferreira",
    "lastAuthor": "Walter Filgueira de Azevedo",
    "authorCount": 2,
    "affiliations": ["Escola de Ciências da Saúde, Pontifícia Universidade Católica do Rio Grande do Sul-PUCRS, Porto Alegre, RS, Brazil."],
    "journal": "Methods in molecular biology (Clifton, N.J.)",
    "journalAbbrev": "Methods Mol Biol",
    "journalCountry": "United States",
    "nlmJournalId": "9214969",
    "issn": null,
    "eissn": "1940-6029",
    "volume": "2053",
    "issue": null,
    "pages": "149-167",
    "citation": "Methods Mol Biol. 2019;2053:149-167. doi: 10.1007/978-1-4939-9752-7_10.",
    "publicationDate": "2019",
    "publicationYear": 2019,
    "epubDate": null,
    "pubmedDate": "2019-08-28",
    "publicationStatus": "ppublish",
    "language": "eng",
    "publicationTypes": ["Journal Article", "Research Support, Non-U.S. Gov't"],
    "isReview": false,
    "meshTerms": ["Binding Sites", "Cyclin-Dependent Kinase 2", "Drug Design", "Humans", "Ligands", "Molecular Docking Simulation", "Molecular Dynamics Simulation", "Protein Binding", "Proteins", "Software", "User-Computer Interface"],
    "meshMajorTopics": ["Molecular Docking Simulation", "Molecular Dynamics Simulation", "Software"],
    "keywords": ["Cyclin-dependent kinase 2", "Drug design", "MolDock", "Molecular docking", "Molegro Virtual Docker", "Protein-ligand interactions"],
    "chemicals": ["Ligands", "Proteins", "Cyclin-Dependent Kinase 2"],
    "fundingAgencies": [],
    "referenceCount": 0,
    "doi": "10.1007/978-1-4939-9752-7_10",
    "pmcid": null,
    "pii": null,
    "medlineStatus": "MEDLINE",
    "recordType": "article",
    "bookTitle": null,
    "bookAccession": null,
    "publisher": null,
    "url": "https://pubmed.ncbi.nlm.nih.gov/31452104/",
    "doiUrl": "https://doi.org/10.1007/978-1-4939-9752-7_10",
    "pmcUrl": null,
    "source": "search",
    "searchTerm": "molecular docking software",
    "position": 1,
    "scrapedAt": "2026-10-20T07:30:12.345Z"
}

A SUMMARY record in the key-value store lists, per search term, how many results PubMed reported, how many were fetched and saved, PMIDs that were not found, request count and any query errors.

Pricing

Pay per event: US$0.55 per 1,000 articles (article-result, US$0.00055 each), plus Apify platform usage, which is small because the Actor is plain HTTP without a browser or proxy.

Run	Articles	Price
Default input (one term, 20 articles)	20	US$0.011
One query, 1,000 articles	1,000	US$0.55
Weekly alert on 5 queries, 200 new papers each	1,000	US$0.55
Systematic review export	10,000	US$5.50

There is no start fee and nothing is charged for duplicates or for PMIDs that do not exist.

How it compares

• NCBI E-utilities directly: free, but you have to handle paging, batching, the 3 requests/second limit, retries on NCBI's transient 500s and the PubMed XML (structured abstracts, MedlineDate, book records). This Actor does that and gives you flat JSON, a dataset you can export as CSV/Excel and an API endpoint.
• PubMed website export: capped and manual; no abstracts in CSV, no automation.
• Other PubMed Actors: most charge US$1–3 per 1,000 records or a per-run start fee, and several scrape the HTML instead of using the official API. This one is API-only, exact maxItems, no start fee.

It does not download full-text PDFs; use pmcUrl (PubMed Central) or doiUrl for that.

Integrations

Run it on a schedule and send the dataset anywhere: Apify API (/v2/datasets/{id}/items?format=csv), webhooks, Google Sheets, Zapier, Make, n8n, Airbyte, or call it from LangChain / LlamaIndex / an MCP client as a tool for AI agents.

FAQ

Is this legal? Yes. PubMed is a public database of the US National Library of Medicine and E-utilities is its documented API. The Actor identifies itself with the tool parameter, keeps under NCBI's rate limit and only carries bibliographic data (author names and affiliations as published; e-mail addresses are stripped).

Do I need a proxy? No. NCBI accepts direct connections; proxyConfiguration is off by default. Turn on Apify Proxy only if you see HTTP 429 in the log.

Do I need an NCBI API key? No. Without a key the Actor runs at 3 requests per second (about 500 articles per minute). A free key from My NCBI raises that to 10 per second.

What are the limits? PubMed returns at most 10,000 results per query; split by date range for more. One article found by several terms is saved once.

How fresh is the data? Live from NCBI at run time, including records "ahead of print".

Changelog

• 0.1 (October 2026): first release. Search terms with full PubMed syntax, PMIDs / URLs, date range, Best Match or newest first, abstracts, authors with affiliations and ORCID, MeSH, keywords, grants, Bookshelf records, optional NCBI API key.

---

En español

PubMed Scraper extrae artículos de PubMed a través de la API oficial E-utilities de NCBI: busca por palabras clave (con la sintaxis completa de PubMed) o por PMID y entrega título, resumen, autores con afiliaciones, revista, fechas ISO, DOI, PMCID, tipos de publicación, términos MeSH y palabras clave como JSON plano. Sin navegador, sin proxy y sin clave de API (opcional para ir a 10 peticiones por segundo). Precio: US$0,55 por cada 1.000 artículos, sin cargo por inicio.