Constat · Stage 2 · Post-deployment

Medical-device postmarket surveillance for FDA-cleared AI.

Name: Constat FDA AI Medical Device Postmarket Surveillance Dataset
Creator: Health AI

Review source-linked MAUDE trends, recalls, re-clearances, warning-letter matches, and disclosed drift signals across FDA's AI-enabled device cohort. Built for postmarket surveillance, quality, and regulatory teams; every signal is an observable, not a safety judgment.

Research method: how to search MAUDE without turning reports into rates

Live cohort — snapshotted AI/ML devices

1,505

devices snapshotted — refreshed weekly, rolling

Latest snapshot 2026-08-19

In a product code with a recall in the last 24 months709 of 1,505

MAUDE reports above the code's own 3-year baseline554 of 808

Of devices where the trend is computable.

Carry a device-specific drift signal745 of 1,505

Own line re-cleared, own applicant's recall, or predicate neighbors.

Carry any signal, incl. code-level1,058 of 1,505

Includes a software recall anywhere in the device's product code.

Applicant matched a recent FDA warning letter26 of 1,505

Of devices scanned while the letters feed was live.

Every figure is a presencerate with its denominator inline — never pooled across devices. Recall and MAUDE counts are product-code-level (MAUDE reports are not reliably attributable to a specific device). Descriptive only: “this code had N software-related recalls” — never “this device is unsafe.”

Drift signals by kind

Software/algorithm recall in the product code668 devices · 6111 signals

Re-clearance of the same device line538 devices · 1624 signals

Adverse-event inflection (MAUDE vs own baseline)308 devices · 308 signals

Recall activity among predicate-sharing devices256 devices · 528 signals

MAUDE database search and limitations

Search adverse-event signals without treating reports as incidence

How do I search the FDA MAUDE database for a medical device?

Search by device name, manufacturer, brand, product code, or report number, then read the underlying MDR records. Constat groups public MAUDE trends with the matching FDA product-code cohort and links those observations to clearance and recall context.

Does a MAUDE report prove that a medical device caused an event?

No. MAUDE is a passive-reporting system with incomplete, duplicate, and unverified reports. A report or count is a signal for review, not proof of causation, incidence, prevalence, or device safety.

What is medical-device postmarket surveillance?

Postmarket surveillance is the ongoing review of evidence after a device reaches the market. Depending on scope, that can include adverse-event reports, recalls, re-clearances, safety communications, warning letters, and other source-linked changes.

Verify records in the FDA MAUDE search.

Machine access

Query the same cohort over MCP — device_postmarket_lookup, postmarket_search, cohort_postmarket_stats

curl -s https://constat.dev/api/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"postmarket_search",
                 "arguments":{"panel":"Radiology","has_drift_signal":true}}}'

The premarket half lives at Precedent — same corpus, same K-numbers, clearance to runtime.

Discuss monitored postmarket surveillance for your device set

Share the device, product code, competitors, and review cadence. Health AI will confirm the source boundary, signal definitions, review process, deliverable, and limitations before monitoring begins.