How we research, source, and review

Sourcing

Every deep dive and digest entry links to authoritative or primary sources: government health bodies (NIH/NIA, CDC, NHLBI, NIDDK, NINDS, NCCIH), major professional organizations (the Alzheimer's Association, American Heart Association), the World Health Organization, MedlinePlus, and peer-reviewed studies via PubMed. We prefer the original study or guideline over secondary coverage.

Accuracy and link integrity

Claims are checked against the cited sources, and an automated test verifies that every citation URL still resolves. We distinguish what is well established from what is emerging or contested, and we state uncertainty plainly rather than rounding it away.

Our evidence vocabulary

Articles open with a short plain-language takeaway, not a study-design label. Internally we still use one of five labels for the strongest study design available for the article's practical question, not the number of papers published or how exciting a result sounds.

Randomized human evidence

Researchers assigned people to an intervention or comparison group. Randomization is the strongest design here for estimating whether an intervention caused an outcome, but duration, adherence, sample, and surrogate endpoints can still limit the conclusion.

Observational human evidence

Researchers measured exposures and outcomes without assigning them. This can show an association in people, including dose patterns and long follow-up, but confounding and reverse causation mean it cannot by itself prove cause and effect.

Early human evidence

A small, short, first-in-human, pilot, or biomarker study has tested the idea in people. It can establish feasibility, safety, or an early signal, but usually cannot establish lasting clinical benefit or rare harms.

Animal or laboratory evidence

The finding comes from animals, cells, tissue, or another laboratory model. It can explain a mechanism and justify human research, but it cannot establish that an intervention benefits people.

Expert hypothesis

A plausible interpretation or recommendation is based on expert synthesis rather than direct outcome evidence. It helps frame a question; it does not demonstrate that the claim is true.

We also note whether APOE4 carriers were directly studied, name the largest limitation, and keep a working sense of how confident the practical conclusion is. A strong general result is not automatically a carrier-specific result, and a carrier subgroup is not automatically large enough to be decisive. That judgment shows up in the prose, not as a confidence badge.

Medical review

This site is general, evidence-based education, not medical advice, so it does not carry a "medically reviewed by" credential. We are honest about that on purpose: claiming a clinical review that did not happen would be misleading. What we lean on instead is rigorous sourcing, clearly stated uncertainty, and prompt corrections. If we ever engage a named clinician with a dated, logged review process, we will say so here and on the relevant articles, and explain exactly what that review does and does not cover.

How AI is used

We use AI tools to help draft and edit, always with human review, source-checking, and editing before anything is published. AI is a writing aid, not the author of record, and it does not replace sourcing or human judgment.

Corrections

If we get something wrong, we fix it and note material changes. Spotted an error or a dead link? Tell us via the About page and we will look into it.

Not medical advice

Everything here is general education, not personalized medical advice. Decisions about your health belong with a qualified clinician who knows your full history.