# PubMed Scraper for RAG: Papers as Chunked JSON (`devanshlive/pubmed-rag-extractor`) Actor Scrape PubMed citations by search term, MeSH, and article type. Returns RAG-ready JSON with full-text chunks from PMC Open Access (cl100k\_base, 512/50) and abstract fallback. Drop-in for LangChain, LlamaIndex, Qdrant, Pinecone, Weaviate, pgvector. Skip GROBID / Pubmed Parser. $0.02 per paper. - **URL**: https://apify.com/devanshlive/pubmed-rag-extractor.md - **Developed by:** [Devansh Tiwari](https://apify.com/devanshlive) (community) - **Categories:** AI, Developer tools, Agents - **Stats:** 2 total users, 1 monthly users, 100.0% runs succeeded, NaN bookmarks - **User rating**: No ratings yet ## Pricing from $20.00 / 1,000 papers This Actor is paid per event. You are not charged for the Apify platform usage, but only a fixed price for specific events. Learn more: https://docs.apify.com/platform/actors/running/actors-in-store#pay-per-event ## What's an Apify Actor? Actors are a software tools running on the Apify platform, for all kinds of web data extraction and automation use cases. In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours, and optionally produces a well-defined JSON output, datasets with results, or files in key-value store. In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server. Actors are written with capital "A". ## How to integrate an Actor? If asked about integration, you help developers integrate Actors into their projects. You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready. The best way to integrate Actors is as follows. In JavaScript/TypeScript projects, use official [JavaScript/TypeScript client](https://docs.apify.com/api/client/js.md): ```bash npm install apify-client ``` In Python projects, use official [Python client library](https://docs.apify.com/api/client/python.md): ```bash pip install apify-client ``` In shell scripts, use [Apify CLI](https://docs.apify.com/cli/docs.md): ````bash # MacOS / Linux curl -fsSL https://apify.com/install-cli.sh | bash # Windows irm https://apify.com/install-cli.ps1 | iex ```bash In AI frameworks, you might use the [Apify MCP server](https://docs.apify.com/platform/integrations/mcp.md). If your project is in a different language, use the [REST API](https://docs.apify.com/api/v2.md). For usage examples, see the [API](#api) section below. For more details, see Apify documentation as [Markdown index](https://docs.apify.com/llms.txt) and [Markdown full-text](https://docs.apify.com/llms-full.txt). # README ## PubMed Scraper for RAG: Papers as Chunked JSON **Scrape PubMed biomedical research into RAG-ready JSON in one call.** Pulls papers by free-text search, MeSH ontology, article type, and date range. Fetches PMC Open Access full-text when available and falls back to the abstract otherwise. Returns fixed-token chunks (cl100k_base, 512 tokens / 50 overlap) with full metadata, ready to drop into **LangChain, LlamaIndex, Qdrant, Pinecone, Weaviate, pgvector, or Chroma**. Built for AI training data teams, pharma/biotech researchers, and anyone who's written "just one more PubMed Parser patch" at 2 AM. ### What does PubMed Scraper for RAG do? This Apify Actor scrapes **[PubMed](https://pubmed.ncbi.nlm.nih.gov/)** citations matching your search criteria, fetches PMC Open Access full-text where available, and splits the resulting plain text into **tokenizer-aware chunks (512 tokens, 50-token overlap, tiktoken cl100k_base)** ready to embed or feed into a RAG index. Each output record contains clean metadata (PMID, PMCID, DOI, title, authors, journal, MeSH terms, article types, publication date) and a `chunks` array of `{ idx, text, tokens }` ready for direct ingestion into a vector database. **Try it in the Apify Console.** Fill in a search term (e.g. `SARS-CoV-2 vaccines`), optional MeSH terms or article types, a date range, a paper cap, and hit Start. Download results as JSON, CSV, or Excel. Built on the Apify platform, you also get: scheduled runs, HTTP API access, integrations with Zapier / n8n / Make, proxy rotation, monitoring, and alerts. No infrastructure to run yourself. ### Why use PubMed Scraper for RAG? - **Skip the PubMed XML parsing grind.** Clean Paper records, not raw E-utilities `` trees. - **PMC Open Access full-text when it exists.** Around 20-30% of PubMed articles are in PMC OA; this Actor grabs NXML and strips it to readable prose. Abstract fallback otherwise. - **MeSH + article-type filtering built in.** Search by MeSH ontology (e.g. `Neoplasms`) or limit to `Review` / `Meta-Analysis` / `Clinical Trial`. - **Pre-chunked for RAG.** `tiktoken cl100k_base` tokenization, compatible with OpenAI `text-embedding-3`, Claude, Cohere, and most BGE/E5/nomic embedding models. - **Vector-DB neutral.** Drop into **Qdrant, Pinecone, Weaviate, pgvector (Supabase / Neon), Chroma, Milvus** without reformatting. - **Framework-ready.** Works with **LangChain, LlamaIndex, Haystack, LangGraph**. - **Respects NCBI rate limits.** 3 requests per second without an API key, 10 per second with one. - **Cheap.** $0.02 per paper. A week of clinical-trial reviews (~200 papers) costs around $4. ### How to use PubMed Scraper for RAG 1. **Open the Actor** in Apify Console. 2. **Set `searchTerm`** (free-text query; supports PubMed syntax like `SARS-CoV-2[mesh] AND vaccine[ti]`). 3. **(Optional) Set `meshTerms`** to filter by Medical Subject Headings. 4. **(Optional) Set `articleTypes`** to narrow to `clinical_trial` / `review` / `meta_analysis` / `case_report` / `observational_study`. 5. **Set `dateFrom` / `dateTo`** in `YYYY-MM-DD` format. 6. **(Optional) Set `ncbiApiKey`** for 10 req/s (free from NCBI). Without it you get 3 req/s. 7. **Click Start.** Expect ~3 sec per paper without an API key, ~1 sec with. 8. **Download results** from the Storage tab. ### Input | Field | Type | Required | Default | Description | |---|---|---|---|---| | `searchTerm` | string | | `SARS-CoV-2 vaccines` | Free-text PubMed query, supports `[mh]`, `[ti]`, `[pt]` syntax | | `meshTerms` | string[] | | `[]` | MeSH headings. Each entry wrapped with `[mh]`, OR-combined | | `articleTypes` | string[] | | `[]` | Enum values: clinical_trial / review / meta_analysis / case_report / observational_study | | `dateFrom` | string (YYYY-MM-DD) | | `2024-01-01` | Inclusive publication-date lower bound | | `dateTo` | string (YYYY-MM-DD) | | `2024-01-15` | Inclusive publication-date upper bound | | `maxPapers` | integer | | `50` | Hard cap on papers returned (1 to 100000) | | `ncbiApiKey` | string (secret) | | | Optional NCBI API key for 10 req/s instead of 3 | **Example input:** ```json { "searchTerm": "SARS-CoV-2 vaccines", "articleTypes": ["review", "meta_analysis"], "dateFrom": "2024-01-01", "dateTo": "2024-01-31", "maxPapers": 100 } ```` ### Output Each paper becomes one dataset item. You can download the dataset in JSON, HTML, CSV, or Excel. ```json { "pmid": "38012345", "pmcid": "PMC10123456", "doi": "10.1038/s41586-024-01234-5", "title": "Durable protection against SARS-CoV-2 variants after boosting", "abstract": "...", "authors": ["Jane Smith", "John Doe"], "journal": "Nature", "mesh_terms": ["SARS-CoV-2", "Vaccines", "Immunization, Secondary"], "article_types": ["Journal Article", "Research Support, N.I.H., Extramural"], "publication_date": "2024-03-15", "pubmed_url": "https://pubmed.ncbi.nlm.nih.gov/38012345", "pdf_url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10123456/pdf/", "source": "full_text", "chunks": [ { "idx": 0, "text": "...", "tokens": 487 }, { "idx": 1, "text": "...", "tokens": 512 } ] } ``` `pmcid`, `doi`, `pdf_url` may be null. `source` is `"full_text"` when PMC OA NXML was parsed, `"abstract"` when it fell back. #### Data table | Field | Type | Description | |---|---|---| | `pmid` | string | PubMed ID (primary identifier) | | `pmcid` | string? | PubMed Central ID (present for PMC OA papers) | | `doi` | string? | DOI (when provided by PubMed) | | `title` | string | Paper title | | `abstract` | string | Abstract as returned by E-utilities | | `authors` | string\[] | Author display names, order preserved | | `journal` | string | Journal title | | `mesh_terms` | string\[] | MeSH Descriptor headings assigned by NLM | | `article_types` | string\[] | PubMed publication types | | `publication_date` | ISO date | `YYYY-MM-DD` | | `pubmed_url` | string | Link to the PubMed citation | | `pdf_url` | string? | PMC PDF link, when available | | `source` | `"full_text"` | `"abstract"` | Text origin | | `chunks` | Chunk\[] | Token-aware chunks for RAG | | `chunks[].idx` | number | 0-indexed position | | `chunks[].text` | string | Chunk text | | `chunks[].tokens` | number | Token count (≤ 512) | ### Pricing **$0.02 per paper** (PPR, pay per result). #### How much does it cost to scrape PubMed? | Volume | Estimated cost | |---|---| | 100 papers | **~$2** | | 1,000 papers | **~$20** | | 10,000 papers | **~$200** | | 100,000 papers | **~$2,000** | No subscription. No minimum. You pay only for successful records. ### Tips - **Get an NCBI API key** (free at https://www.ncbi.nlm.nih.gov/account/settings/) for 10 req/s. Turns a 55-min run into roughly 17 min for 1000 papers. - **Use specific `searchTerm` syntax.** PubMed query operators like `[ti]`, `[mh]`, `[au]`, `[dp]` work inside the `searchTerm` field. Example: `cancer[mh] AND immunotherapy[ti] AND 2024[dp]`. - **Filter article types** to cut noise. `["review", "meta_analysis"]` removes editorials, letters, and news pieces. - **`source: "abstract"` is expected for most records.** Only PMC OA papers have full-text NXML. Plan your RAG index accordingly (abstracts are dense and still make good chunks). - **Schedule weekly runs** to keep an embedding index fresh on newly-published biomedical research. - **For large backfills**, split into month-sized windows and run in parallel (separate Actor runs) to stay under the rate limit cleanly. ### FAQ and limitations #### Is scraping PubMed legal? PubMed's E-utilities API is explicitly designed for programmatic access ([E-utilities docs](https://www.ncbi.nlm.nih.gov/books/NBK25501/)). This Actor respects NCBI's stated rate limits (3 req/s without API key, 10 with) and fetches only publicly-available content. Full-text comes from the PMC Open Access subset, which is licensed for reuse under CC-BY or equivalent. #### What's not in v1? - **Other databases** (bioRxiv, medRxiv, ClinicalTrials.gov, Semantic Scholar). arXiv has its own companion Actor in this portfolio. - **Species / affiliation / journal / language filters** are deferred to a v2 premium tier. - **Citation graph** extraction. - **PDF OCR** for non-PMC papers (they fall back to abstract). - **Section-aware splitting** (Abstract / Methods / Results as separate records). - **MeSH hierarchy expansion** (searching `Cancer` does not auto-include child terms). - **Real-time streaming** or incremental sync. #### Rate limits - **Without API key:** 3 req/s (~55 min for 1000 papers) - **With API key:** 10 req/s (~17 min for 1000 papers) Get a free key at https://www.ncbi.nlm.nih.gov/account/settings/. #### Support Found a bug or want a feature? Use the **Issues** tab on the Actor's page. Custom requirements (other databases, different chunking, section-aware splitting)? Reach out via the Actor's Support link. #### Disclaimer Output metadata is from PubMed's public E-utilities API. Full text, when available, comes from the PMC Open Access archive, which is licensed under CC-BY or equivalent open licenses. Check individual papers' licenses on their PMC page before downstream commercial use. *** **Built with Apify + Crawlee + TypeScript.** Part of the [actorstack](https://github.com/Devansh-365/actorstack) portfolio. Sister Actor: [arXiv Scraper for RAG](https://apify.com/devanshlive/arxiv-rag-extractor). # Actor input Schema ## `searchTerm` (type: `string`): Free-text PubMed query. Supports PubMed syntax (e.g. 'SARS-CoV-2\[mesh] AND vaccine\[ti]'). Empty = no keyword filter (date / MeSH / article type still apply). ## `meshTerms` (type: `array`): Medical Subject Headings. Each entry is wrapped with \[mh] and OR-combined. ## `articleTypes` (type: `array`): Filter by PubMed publication type. Empty = all types. ## `dateFrom` (type: `string`): Inclusive lower bound of publication date. ## `dateTo` (type: `string`): Inclusive upper bound of publication date. ## `maxPapers` (type: `integer`): Hard cap on returned papers. Unauthenticated rate limit is 3 req/s (~55 min per 1000). With an NCBI API key: 10 req/s (~1.7 hrs per 1000). ## `ncbiApiKey` (type: `string`): Free NCBI API key raises the rate limit from 3 to 10 requests per second. Get one at https://www.ncbi.nlm.nih.gov/account/settings/ ## Actor input object example ```json { "searchTerm": "SARS-CoV-2 vaccines", "meshTerms": [], "articleTypes": [], "dateFrom": "2024-01-01", "dateTo": "2024-01-15", "maxPapers": 50 } ``` # Actor output Schema ## `dataset` (type: `string`): No description # API You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup. ## JavaScript example ```javascript import { ApifyClient } from 'apify-client'; // Initialize the ApifyClient with your Apify API token // Replace the '' with your token const client = new ApifyClient({ token: '', }); // Prepare Actor input const input = { "searchTerm": "SARS-CoV-2 vaccines", "meshTerms": [], "articleTypes": [], "dateFrom": "2024-01-01", "dateTo": "2024-01-15", "maxPapers": 50 }; // Run the Actor and wait for it to finish const run = await client.actor("devanshlive/pubmed-rag-extractor").call(input); // Fetch and print Actor results from the run's dataset (if any) console.log('Results from dataset'); console.log(`💾 Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`); const { items } = await client.dataset(run.defaultDatasetId).listItems(); items.forEach((item) => { console.dir(item); }); // 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/js/docs ``` ## Python example ```python from apify_client import ApifyClient # Initialize the ApifyClient with your Apify API token # Replace '' with your token. client = ApifyClient("") # Prepare the Actor input run_input = { "searchTerm": "SARS-CoV-2 vaccines", "meshTerms": [], "articleTypes": [], "dateFrom": "2024-01-01", "dateTo": "2024-01-15", "maxPapers": 50, } # Run the Actor and wait for it to finish run = client.actor("devanshlive/pubmed-rag-extractor").call(run_input=run_input) # Fetch and print Actor results from the run's dataset (if there are any) print("💾 Check your data here: https://console.apify.com/storage/datasets/" + run["defaultDatasetId"]) for item in client.dataset(run["defaultDatasetId"]).iterate_items(): print(item) # 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/python/docs/quick-start ``` ## CLI example ```bash echo '{ "searchTerm": "SARS-CoV-2 vaccines", "meshTerms": [], "articleTypes": [], "dateFrom": "2024-01-01", "dateTo": "2024-01-15", "maxPapers": 50 }' | apify call devanshlive/pubmed-rag-extractor --silent --output-dataset ``` ## MCP server setup ```json { "mcpServers": { "apify": { "command": "npx", "args": [ "mcp-remote", "https://mcp.apify.com/?tools=devanshlive/pubmed-rag-extractor", "--header", "Authorization: Bearer " ] } } } ``` ## OpenAPI specification ```json { "openapi": "3.0.1", "info": { "title": "PubMed Scraper for RAG: Papers as Chunked JSON", "description": "Scrape PubMed citations by search term, MeSH, and article type. Returns RAG-ready JSON with full-text chunks from PMC Open Access (cl100k_base, 512/50) and abstract fallback. Drop-in for LangChain, LlamaIndex, Qdrant, Pinecone, Weaviate, pgvector. Skip GROBID / Pubmed Parser. $0.02 per paper.", "version": "0.1", "x-build-id": "dsqLHRcZgZsmLZaaW" }, "servers": [ { "url": "https://api.apify.com/v2" } ], "paths": { "/acts/devanshlive~pubmed-rag-extractor/run-sync-get-dataset-items": { "post": { "operationId": "run-sync-get-dataset-items-devanshlive-pubmed-rag-extractor", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for its completion, and returns Actor's dataset items in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } }, "/acts/devanshlive~pubmed-rag-extractor/runs": { "post": { "operationId": "runs-sync-devanshlive-pubmed-rag-extractor", "x-openai-isConsequential": false, "summary": "Executes an Actor and returns information about the initiated run in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK", "content": { "application/json": { "schema": { "$ref": "#/components/schemas/runsResponseSchema" } } } } } } }, "/acts/devanshlive~pubmed-rag-extractor/run-sync": { "post": { "operationId": "run-sync-devanshlive-pubmed-rag-extractor", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for completion, and returns the OUTPUT from Key-value store in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } } }, "components": { "schemas": { "inputSchema": { "type": "object", "properties": { "searchTerm": { "title": "Search term", "type": "string", "description": "Free-text PubMed query. Supports PubMed syntax (e.g. 'SARS-CoV-2[mesh] AND vaccine[ti]'). Empty = no keyword filter (date / MeSH / article type still apply).", "default": "SARS-CoV-2 vaccines" }, "meshTerms": { "title": "MeSH terms", "type": "array", "description": "Medical Subject Headings. Each entry is wrapped with [mh] and OR-combined.", "items": { "type": "string" }, "default": [] }, "articleTypes": { "title": "Article types", "type": "array", "description": "Filter by PubMed publication type. Empty = all types.", "items": { "type": "string", "enum": [ "clinical_trial", "review", "meta_analysis", "case_report", "observational_study" ] }, "default": [] }, "dateFrom": { "title": "From date (YYYY-MM-DD)", "type": "string", "description": "Inclusive lower bound of publication date.", "default": "2024-01-01" }, "dateTo": { "title": "To date (YYYY-MM-DD)", "type": "string", "description": "Inclusive upper bound of publication date.", "default": "2024-01-15" }, "maxPapers": { "title": "Max papers per run", "minimum": 1, "maximum": 100000, "type": "integer", "description": "Hard cap on returned papers. Unauthenticated rate limit is 3 req/s (~55 min per 1000). With an NCBI API key: 10 req/s (~1.7 hrs per 1000).", "default": 50 }, "ncbiApiKey": { "title": "NCBI API key (optional)", "type": "string", "description": "Free NCBI API key raises the rate limit from 3 to 10 requests per second. Get one at https://www.ncbi.nlm.nih.gov/account/settings/" } } }, "runsResponseSchema": { "type": "object", "properties": { "data": { "type": "object", "properties": { "id": { "type": "string" }, "actId": { "type": "string" }, "userId": { "type": "string" }, "startedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "finishedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "status": { "type": "string", "example": "READY" }, "meta": { "type": "object", "properties": { "origin": { "type": "string", "example": "API" }, "userAgent": { "type": "string" } } }, "stats": { "type": "object", "properties": { "inputBodyLen": { "type": "integer", "example": 2000 }, "rebootCount": { "type": "integer", "example": 0 }, "restartCount": { "type": "integer", "example": 0 }, "resurrectCount": { "type": "integer", "example": 0 }, "computeUnits": { "type": "integer", "example": 0 } } }, "options": { "type": "object", "properties": { "build": { "type": "string", "example": "latest" }, "timeoutSecs": { "type": "integer", "example": 300 }, "memoryMbytes": { "type": "integer", "example": 1024 }, "diskMbytes": { "type": "integer", "example": 2048 } } }, "buildId": { "type": "string" }, "defaultKeyValueStoreId": { "type": "string" }, "defaultDatasetId": { "type": "string" }, "defaultRequestQueueId": { "type": "string" }, "buildNumber": { "type": "string", "example": "1.0.0" }, "containerUrl": { "type": "string" }, "usage": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "integer", "example": 1 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 } } }, "usageTotalUsd": { "type": "number", "example": 0.00005 }, "usageUsd": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "number", "example": 0.00005 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 } } } } } } } } } } ```