How we organize research evidence

Candidate research is found using ingredient names and major aliases. Not every candidate becomes part of a Goal Tier; unclear ingredient attribution or goal relevance is excluded or kept as separate context.

Studies are grouped as directly examining the goal, involving a specific condition or treatment setting, providing indirect evidence, or excluded from effect evidence. Animal and cell findings are kept separate from human studies.

Goal Tier, where displayed, is an experimental summary of automatically classified research. It is not a validated clinical certainty grade or a professional guideline. Automated analysis can misclassify comparisons, populations, and outcomes, so a tier alone cannot establish efficacy or applicability to a general population.

Counts describe the material currently collected and classified. They are not independent trial counts, quality ratings, or proof of a complete search. Original studies and reviews have different roles. A reported null comparison differs from an unreadable result, and a positive-paper ratio is not a personal probability.

Research intake is an observed range from verifiable Core studies. It is not a recommended dose or an upper safety limit, and many studies do not report a verifiable intake in the available source.

Side-effect and combination information is a literature signal kept separate from Goal Tier calculations. It is not a comprehensive safety grade, and the absence of a displayed signal does not prove safety.

Technical terms are translated into user-friendly language without adding conclusions that were not present in the papers. We prefer wording such as “studies reported this finding” over “this works.”

Available publication metadata and stored abstracts are processed automatically to classify and explain research. This is not a systematic review of full texts. Automated checks that match a source phrase to a measured endpoint do not establish that a person verified the meaning.

We do not claim to have checked every full text, completed a comprehensive risk-of-bias assessment, or had a clinician review each paper. Automated output is limited by the information available in the stored source records.

Use the contact page to report a page URL and the exact statement that needs correction. Information identifying the research can help when a source check is needed. We do not display invented expert credentials or clinical-review badges.