FDA 510(k) Clearances Scraper + New Device Clearance Alerts

Get every FDA 510(k) medical device clearance as clean JSON, CSV or Excel. Data comes from the official openFDA device/510k API (no login, no API key), so it has no anti-bot breakage. Filter by product code, applicant/company, device name keyword, review panel, country, clearance type and decision date range. Diff mode returns only clearances that are new since your last run, which makes it a ready-made FDA clearance alert / monitor.

Keywords: FDA 510k scraper, 510(k) database, FDA device clearances, premarket notification, medical device approvals, openFDA API, K number lookup, FDA product code, De Novo, competitor monitoring, medtech intelligence, regulatory intelligence.

Use cases

• Medtech competitive intelligence: get an alert when a competitor (e.g. Medtronic, Siemens) or a product code (e.g. QIH AI radiology software) gets a new clearance.
• Sales and BD lead lists: newly cleared device makers need distributors, CROs, reimbursement and regulatory consultants, contract manufacturers and marketing. Each record includes the applicant's address and contact.
• Regulatory consultants: benchmark review times (reviewDays), find predicate devices by product code, and track third-party review usage.
• Investors and analysts: track clearance velocity by specialty, country or company.
• AI and RAG pipelines: a structured, current 510(k) feed with links to the summary PDF.

Input examples

{ "productCodes": ["QIH", "LLZ"], "dateFrom": "2026-01-01" }
{ "applicants": ["Medtronic"], "lookbackDays": 60 }
{ "mode": "diff", "advisoryCommittees": ["CV"], "lookbackDays": 90, "stateKey": "cardio-watch" }

Filters within the same field are combined with OR, and different fields are combined with AND. Results are sorted by newest decision first.

Output (one item per clearance)

kNumber, deviceName, applicant, contact, address, city, state, postalCode, countryCode, dateReceived, decisionDate, reviewDays, decisionCode, decision, clearanceType, productCode, deviceClass, regulationNumber, medicalSpecialty, reviewAdvisoryCommittee, statementOrSummary, thirdPartyReview, expeditedReview, fdaDatabaseUrl, summaryPdfUrl, source (+ changeType, firstSeenAt, baselineRun in diff mode).

Example (real record, Oct 2026 test run):

[
  {
    "kNumber": "K260945",
    "deviceName": "Ultima System (UL-800);Ultimate Face Package (UL-800-Face-Set);Ultimate Shape Extension Kit 1 (UL-800-Shape-Ext-Kit-1);Ultimate Shape Extension Kit 2 (UL-800-Shape-Ext-Kit-2)",
    "applicant": "Fit-Pro KFT (Ltd.) - DBA E-Fit",
    "contact": "Janos Papp",
    "address": "Szabadság Ter. 7",
    "city": "Budapest",
    "state": null,
    "postalCode": "1054",
    "countryCode": "HU",
    "dateReceived": "2026-03-23",
    "decisionDate": "2026-09-27",
    "reviewDays": 188,
    "decisionCode": "SESE",
    "decision": "Substantially Equivalent",
    "clearanceType": "Traditional",
    "productCode": null,
    "deviceClass": null,
    "regulationNumber": null,
    "medicalSpecialty": null,
    "reviewAdvisoryCommittee": "PM",
    "statementOrSummary": "Summary",
    "thirdPartyReview": false,
    "expeditedReview": false,
    "fdaDatabaseUrl": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K260945",
    "summaryPdfUrl": "https://www.accessdata.fda.gov/cdrh_docs/pdf26/K260945.pdf",
    "changeType": "new",
    "firstSeenAt": "2026-10-07T13:46:21+00:00",
    "baselineRun": false
  }
]

"New since last run" (diff mode)

Set Mode = diff and schedule the actor daily or weekly. Each run looks at a rolling date window and returns only records it has never returned before, tagged changeType: "new" and firstSeenAt. Seen IDs are stored in a named key-value store (stateStoreName, separate stateKey per saved search), so the state survives between scheduled runs. On the first run you can output the whole window (flagged baselineRun: true) or store a silent baseline (firstRunBehaviour: baselineOnly). Pair it with an Apify integration (Slack, email, Google Sheets, Zapier, Make, webhook) to get a lead alert feed.

Tip: openFDA refreshes weekly and back-fills decisions, so use a window of 60 to 90 days in diff mode. Late-arriving clearances are still caught, and duplicates never come through.

Limits

• Keyless openFDA allows 1,000 requests/day per IP (each request returns 1,000 rows). You can add an optional apiKey for more.
• openFDA paging stops at 25,000 rows per query, so split very large historical pulls by date range.
• openFDA lags the FDA website by roughly 1 to 2 weeks.

Source and licence

Data from openFDA (U.S. Food and Drug Administration), public domain / CC0 (https://open.fda.gov/license/). Every item carries a source attribution block. This actor is not affiliated with or endorsed by FDA. openFDA's disclaimer applies: do not rely on openFDA for medical care decisions, and treat results as unvalidated.

Row cap and failure handling

• maxRowsPerSource (default 5000, 0 = unlimited): per-source cap on raw rows fetched, with automatic pagination. Sources that hit the cap are listed under cappedSources in the OUTPUT record and in the run's status message.
• Failures are loud: if any source fails (HTTP error, ArcGIS error, timeout), the run is marked FAILED, the status message names the failed source(s) and error, and the OUTPUT key-value record contains ok: false and failedSources. Rows from a failed source are never returned partially.