Methodology
The platform is designed around traceability: source records first, normalized fields second, AI analysis clearly marked, and patent/IP intelligence separated by confidence and source status.
Trial reports start from public registry records, primarily ClinicalTrials.gov. Source identifiers, source URLs, source update dates and internal sync dates are shown wherever available so users can trace important claims back to the registry.
Imported source fields are preserved separately from normalized fields. Conditions, indications, modalities, targets, interventions and sponsors are treated as distinct concepts to avoid presenting disease names as therapeutic modalities.
AI is used to help structure summaries, normalize trial concepts and produce analyst-facing research notes. AI-generated content is marked as analysis and should be validated against source records, publications, sponsor disclosures and domain experts.
Patent rows are shown only when a concrete publication or patent code is available from a structured or curated source. Broader asset and indication landscape records are separated from trial-specific patent records and are research intelligence only.
Public pages are intended to be indexed only when they have enough source-backed content and validated taxonomy. Incomplete records may remain searchable inside the product while being excluded from public SEO surfaces.
TrialSignal is not a medical, legal, regulatory, investment, patentability or freedom-to-operate opinion. Users should validate important findings with primary sources and qualified professionals before relying on them.
Data transparency
Source record
Registry-derived data and source links.
AI-normalized
Structured fields generated from source values.
AI analysis
Research notes and summaries requiring validation.
Human reviewed
Shown only when a persistent review state exists.