Constat · Stage 1 · Premarket evidence

The evidence FDA actually accepted.

Every parsed AI/ML 510(k) summary as structured, source-quoted fields — validation design, endpoints, predicate chains — connected to what happened after clearance and how devices got paid.

Presence rates, never pooled.Every figure below is “reported in X of Y audited devices.” Performance values are never averaged across devices — each stays with its analysis unit, task, and a verbatim FDA quote. A device we haven’t parsed yet is queue, never “FDA accepted thin evidence.”

What evidence did FDA accept for devices like yours?

Keyword-based retrieval over the parsed corpus (names, algorithm descriptions, endpoints, source quotes) — not semantic search. Try: · ·

Corpus coverage — Radiology

1,143 of 1,146clearances parsed, most-recent-first · 3 in parse queue · latest decision 2026-03-30

The hatched segment is our parse queue, not device data. Queued devices never count in any rate below — every denominator is audited devices only.

Original findings from the corpus

129 days

Median FDA review time, received → decision

Across all 1,146 matched AI/ML 510(k)s; the slowest decile takes 8.5+ months (p90 ≈ 260 days). Computed from openFDA decision dates — not published anywhere else for this cohort.

8.7%

PCCP adoption in 2025 AI clearances

Up from 0% before 2023 (5.0% in 2024). Nine in ten AI/ML clearances still file without a Predetermined Change Control Plan — a live gap as FDA pushes the mechanism.

What the audited corpus reports

Evidence reporting

Any sensitivity metric284 of 1,143

Canonicalized — includes per-finding sensitivities, not just the summary's top-line slot.

Sensitivity and specificity236 of 1,143
Any performance metric600 of 1,143
Clinical (not bench-only) data822 of 1,143
Subgroup / demographic reporting645 of 1,143
Predetermined Change Control Plan37 of 1,143

“Not reported” is a real finding about what FDA accepted — the summary was audited and does not state it.

Metric presence (canonicalized)

  • sensitivity284 of 1143
  • specificity239 of 1143
  • auroc163 of 1143
  • ppv72 of 1143
  • npv36 of 1143
  • accuracy125 of 1143
  • f125 of 1143
  • dice236 of 1143
  • iou10 of 1143
  • detection_rate5 of 1143

Presence only — values are never pooled or averaged across devices. Each measurement lives on its device's page with its analysis unit and verbatim quote.

Median predicate age: 1.9y across 1427 of 2061 predicate edges (69% datable, dates backfilled from openFDA) — how old each cited predicate was when the child device cleared.

Study design & generalizability

Where FDA review is tightening — and what almost no premarket dataset reports. Each rate is presence over audited devices; a device that describes no clinical study is honestly “not reported” here.

Prospective or reader (MRMC) study197 of 1,143

Primary validation is prospective-clinical or a multi-reader design — not a retrospective bench run.

Multi-site validation509 of 1,143
External / out-of-distribution testing602 of 1,143

Validation data explicitly independent of the training/development set.

Scanner / device diversity595 of 1,143

Tested across more than one scanner or device manufacturer.

Subgroup performance broken out580 of 1,143

Performance reported per subgroup, not just an overall number.

Extraction quality. Every field carries a verbatim FDA quote. Across all 1,146 parsed devices we check every one of ~20,000 quotes against its source 510(k): 98.4% grounded (99% mean token recall; a reproducible, published method). Two records are also hand-audited against the full documents at 100% field accuracy, including a deliberate no-new-testing device to catch invented metrics. We measure fabrication, not interpretation — and publish the method.

One corpus, whole lifecycle

Every device page connects all four stages: clearance → evidence → postmarket → payment. Stages without data render as “not yet tracked”, never as empty or fabricated.

Machine access

The same retrieval over MCP — 12 tools across the lifecycle
curl -s https://constat.dev/api/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"evidence_search",
                 "arguments":{"panel":"Radiology","reports_any_sensitivity_metric":true}}}'

device_evidence_lookup, evidence_search, predicate_chain, device_postmarket_lookup, reimbursement_lookup, and more. Every extracted field carries a source quote and page — descriptive only, never a compliance judgment.

Connect the MCP endpoint

Follow the corpus as it grows

New parses land weekly, most-recent-first. Get notified as coverage expands and presence rates shift — and be first on the client digest when it ships.

Constat Precedent, the premarket evidence layer, by Health AI. Built on public FDA data (openFDA, accessdata.fda.gov); descriptive decision-support, not regulatory advice or consulting.