Skip to main content
How we research

Our Evidence Standard

Where the facts on this site come from, and exactly how we decide how much weight to give them. This page is about the science: sourcing, evidence hierarchy, and how a confidence rating actually gets computed.

Most health writing online has a problem. It either tells you what to do with total confidence it has not earned, or it buries one real finding under a thousand words of padding. We are trying to do something narrower and more honest than that.

Someone asks a question. Does creatine cause hair loss. Is a continuous glucose monitor worth it if you are not diabetic. What does my resting heart rate actually mean. We did not always know the answer either. So we went and read the studies, and this is where we tell you exactly how we do that, so you can decide how much to trust what you read here.

This page is about the science: sourcing and confidence ratings. For the actual production process a study goes through on its way to becoming an article, see How KnowYourPrime Works.

What this system does not do

It does not diagnose anything. It does not personalize to your individual data or history. It does not replace a clinician. It does not predict your individual health outcomes. And it does not give advice: it reports what research found, never what to do about it.

What we actually do

Every article starts the same way. A real question that real people are asking. We find what the published research says about it, we read the studies themselves rather than someone's summary of them, and we report what they found in plain language. Then we tell you where the evidence runs out.

Concretely, that process is: every article begins with peer-reviewed research, tracked and rated on our Research Desk (each metric has its own evidence file, linked from that metric's hub page). AI drafts the article from that cited evidence, organizing the studies, writing the explanation, matching the language to how strong the evidence actually is. Before anything publishes, an automated gate checks the draft: every claim must trace to a cited source, language strength must match evidence strength (a correlation can't be written as a cause), and each topic's confidence rating is computed from the underlying studies using the same fixed rules every time, not judged case by case. A draft that fails any of those checks does not publish. It's rewritten or dropped. There is no manual editorial read of every article before it goes live. That's true today, and this page says so plainly rather than implying otherwise.

That last part matters more than it sounds. The honest answer to a lot of health questions is "the research points one way, but it is thinner than the internet makes it sound." We would rather tell you that than pretend to a certainty nobody has.

The one line that governs everything here: we read the research, and we report what we found. We do not tell you what to do with it. That is your decision, ideally with a doctor who knows your situation.

Publishing depends on passing a fixed set of automated checks, not on how confident the AI sounds.

How we handle sources

Every factual claim about research on this site traces back to a real, named source. Not "studies show." Not "experts agree." A specific study, linked, that you can go read yourself. When a claim matters, we name the kind of evidence behind it, because not all evidence carries the same weight.

Here is the order we trust it in, strongest first:

We also read what people are saying in communities, forums, and comment threads. But that tells us what people are asking and experiencing, not what is true. We never treat a popular belief as evidence. We treat it as the question worth answering.

The rules we hold ourselves to

Claims match the evidence

If the research found a correlation, we do not write it as cause. If one study found something, we do not call it settled. If the evidence is mixed, we say it is mixed, and we say so plainly rather than hiding it in a hedge. You should always be able to tell how much weight a finding can actually bear.

Fact and interpretation stay separate

What a study found is one thing. What we think it might mean is another. We keep those visibly apart, so you are never left guessing whether you are reading a result or our reading of it.

We never give advice

You will not find "you should take this" or "stop doing that" here. We report what the research found and what happened to the people in the studies. What you do with your own body, your own medication, your own health, is a conversation for you and your doctor. We are a place to understand the question better before you have that conversation.

We keep it current

Health evidence moves. A finding that looked solid three years ago can shift. Every article carries the date we last checked it against current research, and when new evidence changes the picture, we update the article and the date.

What the evidence badge at the top of every article means

You will see a small badge at the top of every article (Established, Emerging, Uncertain, Insufficient, or Unsupported) linking straight to the specific Research Desk claim behind that article. Every one of these words has one fixed meaning, computed the same way every time from the actual studies behind that claim, never judged case by case:

The badge means every research claim in the piece was checked against the published studies cited on that Research Desk claim, and the strength of what we wrote was matched to that word's meaning above. It is a statement about our sourcing process, verifiable by clicking through to the actual evidence. It is not a doctor's sign-off, and we would never pretend it was one.

We think honest sourcing you can verify yourself is worth more than a credentialed name attached to content that person barely touched. The studies are linked. You can check our work. That is the trust we are offering, and it is the only kind we think is real.

What is and isn't checked

What's tested on every article, automatically, before it can publish: every factual claim links to a real, named, checkable source; language strength matches evidence strength, so a single study's finding can't be written as settled and a correlation can't be written as a cause; anything drawn from a preprint (research posted before peer review) is explicitly labeled as not-yet-peer-reviewed; there are no missing citations.

What this system cannot do: it cannot determine whether a published study itself is correct. It reports what studies found; it does not re-run their statistics or re-verify their methodology from scratch. It cannot detect scientific fraud in a source paper. And it cannot guarantee that new evidence won't overturn a current rating: ratings are dated and revised as research changes, not fixed forever.

Known ways this can go wrong, stated plainly rather than left for you to discover: AI can misread nuanced statistics or oversimplify genuinely mixed evidence. It inherits any errors already present in a source study. It can lag behind very recently published research.

There is no manual editorial read of every article before it publishes. There is a real, narrower human check, though. For exactly how that works and why it's built that way, see How KnowYourPrime Works.

Corrections

When an error is found, in a source, in a claim, in how something was worded, the article is corrected and the date updated. This has happened before and will happen again. A system that never admits a mistake is a system you should trust less, not more.

What we are not

We are not a medical provider. Nothing here is medical advice, a diagnosis, or a treatment plan, and nothing here is a substitute for the doctor who knows your history. Research describes what tends to happen across groups of people. It cannot tell you what is right for you. For that, see someone who can examine you and read your chart.

If an article ever reads like it is telling you what to do, we got it wrong, and we want to know. The whole point of this site is to hand you a clearer version of the question, not to make the decision for you.

A note on who writes this. Articles are written by Kate Maren, KnowYourPrime's editor. She is not a physician or a scientist, and she never claims to be. The expertise on this site lives in the sourcing and the honesty about limits, not in a claim of medical authority she doesn't have.

Last updated: July 2026