How we research
Medcelerator uses AI to research and write treatment information. Our method aims to make evidence traceable and the limits visible. It cannot replace clinical judgement or guarantee that every claim is correct.
Sources and claims
Our research instructions prioritize regulator decisions and prescribing information, national product monographs, clinical-trial registries, and original research papers. A social post is a lead to investigate, not supporting evidence. Company announcements must be distinguished from independently retrieved regulatory records.
We keep a study’s results separate from the uses authorized by a regulator. A trial for a different age group, treatment stage, or combination does not expand a medicine’s approved use. Unknown coverage or eligibility should remain unknown.
Country-specific access
Approval, commercial availability, public or insurance coverage, and trial eligibility are separate questions. An approval in one country does not establish access in another. The access section distinguishes treatment-specific routes from possible routes and general research resources.
Checks before publication
The Studio publication checks look for a complete five-question patient summary, source links, unfinished summary text, internal editorial labels, and missing descriptions for meaningful images. For existing articles, they also check stored treatment-specific routes and article-specific organization roles for selected contradictions. A separate read-only audit flags shared organization descriptions that may need review.
These are automated structural and consistency checks. They do not verify every medical claim, establish reading comprehension, or constitute medical review. AI-generated content still needs careful evaluation against its sources.
Dates and review
A publication date records when an article was published. Where a summary includes a sources-checked date, it applies only to that summary. Editing a page does not mean its medical evidence has been rechecked. We do not infer verification dates from database timestamps.
Research agents prepare unpublished drafts. Publication is managed in Studio. A Studio publication decision is not automatically a clinician’s review, and this site does not claim a named medical review unless one is documented.
Writing and images
New summaries answer what changed, who may benefit, benefits and uncertainties, availability and coverage, and what to ask a clinician. Older articles are being updated to this format. Detailed evidence remains available below the summary.
Images should explain the treatment or show a relevant product, device, or subject. Drug-name title cards, logos, and decorative branded backgrounds do not qualify as article illustrations. Articles can be complete without an image.
Corrections
If something looks wrong, report it to the publisher. Our correction policy is to check the cited evidence and affected page components, then explain material corrections in the article. The presence of a citation alone does not resolve a contradiction.