How TrialClarity works

We separate factual search from explanation, show uncertainty, and keep the original public record one click away.

Source of truth

Trial records come from the ClinicalTrials.gov API v2. Sponsors and investigators submit those records; ClinicalTrials.gov does not independently approve the safety or science of every listed study. We check for current records regularly and show the study identifier, source version, and last posted update.

What we change

We normalize recruitment status, phase, purpose, interventions, conditions, locations, and common lung-cancer terms so they can be searched consistently. Studies can belong to several categories at once, including cancer type, disease setting, biomarker, treatment approach, purpose, phase, design, and patient-centered topics. Structured source fields receive the highest confidence. Derived categories use conservative exact-term rules and retain a short explanation of the evidence. We reorganize sponsor-provided fields for readability but do not alter the underlying source record, and each study links to its original ClinicalTrials.gov page.

How possible matches are produced

The guide turns selected answers into structured filters. A study surfaces when its public record contains the relevant terms. This is intentionally broader than a clinical screening process. A person may appear to match and still be ineligible—or appear not to match because the public record is incomplete.

AI and automation

The AI guide is connected to a bounded trial-search tool. It searches the normalized public records before explaining possible matches and receives only the fields needed to produce the answer. Requests are not used as enrollment decisions, and the guide is instructed not to collect names, contact details, dates of birth, medical records, or precise addresses. When the language-model service is unavailable, the same interface falls back to transparent keyword and location matching.

Ranking and funding

Organic results are never pay-to-rank. Current results are ordered by source response and filter relevance. If sponsored recruitment is added, it will be visually separated, explicitly labeled, based on approved recruitment material, and unable to change organic ordering.

Corrections

For source-record errors, contact the study sponsor through ClinicalTrials.gov. For a TrialClarity categorization, summary, privacy, or accessibility issue, email corrections@lungcancerclinicaltrials.org with the NCT identifier and what appears wrong.

Methodology version 1.2 · Last reviewed August 29, 2026