Clinical Trial Signal Pricing
Methodology

How signals are classified.

Severity labels are operational guides for screening, not clinical or regulatory judgments. They reflect how attention-worthy a public-record change appears in context — nothing more.

Severity labels

Monitor

Worth knowing about

Routine public-record movement. Track but no urgency. Examples: new Phase 1 trial added, routine label edit, expected results posting.
Notable

Above background

Material change in trajectory. Examples: primary completion date moved > 90 days, Class II recall, label warnings section updated, FDA Drug Safety Communication issued.
High attention

Worth investigating today

Status reversal, Class I recall, suspension or termination of an active trial, multiple independent sources converging on the same signal.
Needs review

Ambiguous — analyst should read

Signal does not fit cleanly into the above. May indicate data anomaly, ingestion artifact, or genuinely novel pattern. AI brief should flag rather than score.

What we do — and do not — infer

We do
  • Report what public-record fields changed and when
  • Surface convergence across sources (e.g., FAERS trend + label change + EDGAR disclosure)
  • Tag the severity-screening label per the criteria above
  • Link every claim back to its primary record
  • Include a caveat appropriate to the source's known limitations
We do not
  • Claim causation from spontaneous-report data
  • Calculate incidence rates from public data alone
  • Make treatment, prescribing, or clinical recommendations
  • Replace validated pharmacovigilance systems of record
  • Provide legal or regulatory determinations
  • Paraphrase official agency language away from its primary form

Audit trail

Every signal stored includes: source name, source URL, source-published timestamp, ingestion timestamp, raw source-record reference, normalized output, and caveat. Customers on Enterprise tier get the Evidence Room export — raw + normalized for every signal in a watchlist over any time window.

Methodology is publicly documented because trust in this category requires it. If you find a signal misclassified, email tom@phase3ai.com with the source URL and your reasoning — we treat reclassification requests as research feedback, not customer complaints.