How Supplement AI Generates Evidence Ratings

Supplement AI's evidence rating is a paper-level research quality signal. It helps determine how much weight to place on the outcomes and conclusions of a given study before that study contributes to broader evidence analysis.

Evidence rating is not meant to discredit studies; it is about consistency. Scientific literature is vast and uneven. By applying a uniform set of standards, Supplement AI makes it easier to identify which studies look methodologically strong and which should be interpreted with more caution.

Our Research Database

Our research database includes over 185,000 papers and counting. Traditionally, studies are ranked by relevance or citation metrics alone. Supplement AI goes further by analyzing each paper's methodological rigor, transparency, and reliability along with factors like journal quality and citation influence.

Methodology

To evaluate methodology, we draw on established frameworks in evidence-based research, including principles from Cochrane, NIH/NHLBI, PRISMA, ARRIVE, and CARE guidelines. Each study type—clinical trial, observational study, systematic review, preclinical model, or case report—is assessed within its own methodological context. This ensures our scoring reflects how well a study was designed and executed, not simply what it found.

The 100-Point Scale

Scores are reported on a 100-point scale, where higher scores indicate stronger, more reliable research. A score near zero reflects serious methodological flaws, while a perfect score represents exceptional scientific rigor.

Dynamic and Transparent

Evidence rating is dynamic, evolving as our analytic models improve. Ratings are recalculated for greater precision but always using consistent criteria. If a rating appears inaccurate, users can flag it for reassessment. We never manually change scores, but we continually refine how they're determined.

In short, evidence rating is the narrow research-quality layer. It helps Supplement AI avoid giving the same weight to a careful clinical study and a weaker or less transparent paper.