How We Verify What We Publish

Mammologix publishes about breast imaging quality, follow-up, and regulatory compliance. Readers use this material to think about how their programs run, and in some cases how patients move through them. That places our content in a category search quality evaluators call Your Money or Your Life, meaning content where inaccuracy can affect someone’s health, finances, or safety.

Content in that category deserves a higher standard than an editor’s judgment on a given afternoon. This page describes the standard we hold ourselves to, the process that enforces it, and the things it does not guarantee.

The instrument

Every article intended for peer and executive readers is scored against the Article-Level Experience, Expertise, Authoritativeness, and Trust Instrument, which we call AEI-100. It returns a score out of 100. Our publication threshold is 95.

AEI-100 is our own instrument. It is derived from concepts described in Google’s Search Quality Rater Guidelines, and it is not a Google score, a Page Quality rating, or any form of external certification. No outside body issues it and no outside body validates it. We built it, we run it, and we are accountable for it.

The four dimensions are weighted deliberately rather than evenly.

Trust carries the heaviest weight because the framework we drew from treats it as primary. Content that cannot be trusted is not improved by being expert.

The gates

Eight conditions are evaluated before any scoring begins. Each is binary. A single failure scores the article zero regardless of points earned elsewhere.

An article fails if any health, clinical, regulatory, or performance claim carries no verifiable source. If any reference fails validation. If certainty language overstates or understates the evidence behind it. If a commercial interest bearing on the subject is undisclosed. If there is no named author with accurate credentials. If commentary appears without our disclaimer. If patient-facing content instructs a reader on her care rather than referring her to her own physician. Or if the article functions as marketing rather than as editorial content.

That last gate deserves a plain statement. Mammologix sells services in the field it writes about.That is a real conflict, and pretending otherwise would be the fastest way to forfeit a reader’s trust. Our rule is that our methodology may be the subject of analysis, but it may never be presented as the answer to a problem an article describes. An article that identifies a gap and then positions us as the solution fails the gate and does not publish.

The process

Verification is separated from writing on purpose. The people and systems that produce an article do not decide whether it is good enough.

Source reconnaissance. Before drafting, we verify the official name of every program, tool, or regulation involved, identify which source is controlling, and document where published sources disagree. Facts we believe to be true are treated as claims to be checked, not as established.

Drafting and author review. The article is written and then reviewed by the author, who directs revision.

Independent evaluation. The finished draft is submitted to a separate evaluation with no knowledge of who wrote it, what it was trying to accomplish, or how many revisions preceded it. Evaluation without that separation tends to confirm rather than test.

Threshold. Scoring below 95 returns the article for revision. An article that fails three cycles is not published in that form. It is rescoped, retired, or published with the failing criteria and our reasoning documented. That last option exists, and we use it rarely.

Publication integrity check. After approval, a final check confirms that nothing broke between approval and publication: citation markers intact, links resolving, references correctly ordered, disclosures present.

Reference standards

Every article is fully referenced. References follow the American Medical Association Manual of Style, 11th edition, numbered in order of first citation, with bracketed numerals placed in the text beside the claim each source supports.

Each reference passes six checks. The source exists and is retrievable. Every bibliographic detail matches. The cited source actually supports the specific claim it is attached to, rather than something adjacent to it. The link resolves. The formatting conforms. The numbering matches the order of citation.

The third check is the one most often skipped in published work, and it is the one that matters most. A correctly formatted citation attached to a claim the source does not make is worse than no citation, because it survives inspection.

Where a source sits behind a paywall, we look for an openly accessible version through the publisher, PubMed Central, or the issuing agency. When we find one, we link it. When we cannot, we mark the reference as paywalled rather than substituting a weaker source that happens to be free. A more accessible source never replaces a stronger one.

Where we cannot verify a source in full, we say so rather than reporting it as verified.

What this does not guarantee

An editorial standard is not a clinical one. This process is designed to catch unsupported claims, mismatched citations, overstated certainty, and undisclosed interest. It is not designed to, and does not, do the following.

It does not make our content medical advice. Nothing we publish substitutes for a licensed physician’s judgment about an individual patient.

It does not make us right. A well-sourced article can still be wrong, and a field that changes will eventually outrun anything written about it. Our standard governs how carefully we work, not whether the conclusion holds.

It does not extend to material we did not write. Sources we cite are subject to their own standards, not ours.

And it does not apply retroactively. This system was implemented in August 2026. Articles published before that date were written under an earlier and less formal standard. We are working through the existing library and bringing it into compliance. Until an article carries a review date on or after that implementation, assume it has not been evaluated under this process.

We would rather tell you that than let a policy page imply a guarantee that does not yet cover everything on the site.

Corrections

If you find an error, tell us. Send the article, the specific claim, and what you believe the correct information is, to connect@mammologix.com.

Substantive corrections are made to the article, noted in the text, and reflected in the review date. We do not silently revise. An article that changes in a way that alters its meaning should show that it changed.