Three US health systems publish records that researchers, medtech teams, and safety watchers check again and again: which clinical trials are recruiting and where, which products were just recalled or cleared, and which medical devices were reported in adverse events. All three are public, and all three are built for looking up one study, one firm, or one report at a time.
I built three Apify Actors that turn those feeds into clean, structured rows you can schedule, filter, and pipe into a spreadsheet or database. Disclosure: I built these, AI-assisted, and I sell runs of them on Apify.
1. ClinicalTrials.gov recruiting sites
ClinicalTrials.gov is the US registry of clinical studies. The registry record for one trial can list dozens of recruiting sites, and answering a simple question, like which sites in one state are recruiting for one condition, means opening trial after trial by hand.
The ClinicalTrials.gov Recruiting Sites Actor reads the official ClinicalTrials.gov API v2 and returns one row per recruiting site: sponsor, phase, condition, intervention, facility, and city and state, newest updates first. You can filter by condition, sponsor, phase, country, state, and site status, and published site contacts are optional.
Use it for: site feasibility research, watching which sponsors are recruiting for a condition, and building a current list of active trial sites in a state or metro.
Limits: a row reports what the sponsor registered. Recruiting status changes, and a site listed as recruiting may have paused or finished enrolling. Check the study record on ClinicalTrials.gov before relying on a row.
https://apify.com/groundplanes/clinicaltrials-recruiting-sites
2. FDA recalls and 510(k) clearances
The FDA publishes enforcement reports covering medical device, drug, and food recalls, and it publishes 510(k) premarket notifications, the clearances that let most medical devices reach the US market. Both live in the openFDA API, and both reward watching on a schedule: a recall in your product line or a competitor clearance is news the week it posts, not a quarter later.
The FDA Recalls and 510(k) Clearances Actor pulls both from the public openFDA API, no API key needed. Filter by recall class, status, keyword, firm, state, and date, newest first.
Use it for: regulatory monitoring, a weekly recall watch for products you make, sell, or stock, and medtech competitive intelligence on who just received a clearance.
Limits: a recall row reports what the FDA published in the enforcement report, and recall scope and status change as firms post updates. A 510(k) clearance is a substantial equivalence decision, not an FDA approval. Check the source record before acting on a row.
https://apify.com/groundplanes/openfda-recalls-510k
3. FDA MAUDE device adverse events
MAUDE is the FDA database of medical device adverse event reports: malfunctions, injuries, deaths, patient outcomes, and manufacturer narratives, submitted by manufacturers, facilities, and patients. It is the main public window into how devices behave after they reach the market, and it is also the easiest of the three to misread, which is why the limits below matter.
The FDA MAUDE Adverse Events Actor returns the reports as structured rows from openFDA, with the unverified submissions disclaimer carried into the output.
Use it for: post-market surveillance research, tracking malfunction patterns for a device type over time, and early safety signal research before a formal study exists.
Limits: MAUDE reports are unverified submissions. A report does not establish that the device caused the event, and report counts are not incidence rates, because reporting is voluntary and incomplete. Treat the output as leads for investigation, not as measured risk.
https://apify.com/groundplanes/openfda-maude-adverse-event-scraper
The pattern
All three follow the same rules I use for the rest of this catalog:
- Public source data only, read through the agency feed or API
- Structured JSON out, with counts stated when a run hits a cap (a capped run says so, it does not silently truncate)
- Schedulable, so a weekly check costs minutes instead of an afternoon
- Status reporting that flags a partial result instead of calling it complete
If you check any of these three feeds by hand today, one scheduled run is the cheapest test of whether automating it is worth it.
Found a problem or need a field? Open an issue on the Issues tab of the Actor page; we respond quickly.


