Editorial Policy
How TrialsAlert sources, drafts, reviews, and updates its content. Last updated 23 April 2026.
How our content is generated
TrialsAlert content falls into three buckets: patient guides (long-form explainers written by the editorial team), condition pages and weekly blog posts (AI-drafted from ClinicalTrials.gov source data and reviewed before publication), and per-subscriber weekly briefings (AI-drafted for each user and delivered by email). In every case, the underlying source of truth is ClinicalTrials.gov. The AI does not invent trial data. It summarizes, classifies, and translates what is already registered there.
Data sources
All clinical trial data comes from the ClinicalTrials.gov API v2, maintained by the U.S. National Library of Medicine at the National Institutes of Health. We fetch updates daily. When we cite trial phases, sponsors, NCT IDs, or eligibility criteria, those figures come directly from ClinicalTrials.gov. When we cite higher-level guidance (for example Phase 1 participant counts or regulatory pathways), we reference the FDA, NIH, and peer-reviewed clinical trial literature.
Update frequency
- Daily. The pipeline fetches new and updated trials from ClinicalTrials.gov, scores them with AI, and enqueues alerts for matching subscribers.
- Weekly (Friday). Each subscriber receives a bundled digest of the trials that matched their tracked conditions during the week.
- Weekly (per disease domain). One public blog post per disease domain summarizes the week's notable trials.
- On publication. Condition page summaries refresh when a new weekly summary is available for that condition.
AI disclosure
Blog posts, condition summaries, weekly digests, doctor report drafts, and trial scores are generated using artificial intelligence. The model currently in use is OpenAI gpt-4.1-mini. AI output is validated programmatically (field checks, banned-word list, numeric sanity checks) and reviewed by the editorial team before being shown to subscribers or indexed for public pages. AI is not used to produce medical advice or replace clinical judgment. We use hedged language by design. We avoid hype words such as breakthrough, miracle, cure, revolutionary, and game-changer, and we never present an AI summary as a substitute for a conversation with a qualified healthcare provider.
Editorial review
Every new template (guide, comparison page, or blog post schema) is reviewed by the founding editorial team before going live. AI-drafted output is sampled weekly for accuracy against the underlying ClinicalTrials.gov records. When the pipeline detects parsing errors, they are logged and the trial is excluded from user-facing surfaces until the error is resolved.
Corrections policy
If you spot an inaccuracy on any TrialsAlert page or in a briefing email, please report it to support@trialsalert.com. We acknowledge corrections within two business days, fix confirmed errors within five business days, and add a dated correction note when the correction affects a published blog post. If a correction reflects a change in the underlying trial record on ClinicalTrials.gov, we link to the updated trial record.
Conflicts of interest
TrialsAlert is an independent subscription service. We do not conduct clinical trials, recruit participants, or have any financial relationship with trial sponsors. On comparison pages (/compare/*) we explicitly disclose that TrialsAlert is our product.
Not medical advice
TrialsAlert is an informational and monitoring service. Nothing on this site is a substitute for advice from a qualified healthcare provider. See our Medical Disclaimer for the full notice.