How Healthcare Discovery scores the gap between what a health claim promises and what the evidence has earned.
Hype is not the absence of evidence. It is a calibration failure: the measurable distance between an entity’s strongest public claim and the level of confidence its evidence actually supports at its stage of development. A laboratory describing preclinical results as preclinical results is calibrated, however thin the data. A commercial product with real trials behind it that is sold beyond what those trials showed is hype, however thick the data. The HYPE Index measures that distance on a common rubric across companies, products, influencers, research papers and, on this site, marketed claims about supplements. It is the methodology behind every evidence grade and every HYPE score published on Healthcare Discovery, developed under the Diligence Dx framework in the HDAI Methodology Series.
Why the gap, not the pile
Most evidence-grading systems count studies. That approach has a blind spot: it rewards quantity over calibration, and it cannot distinguish an honest early-stage claim from a dishonest late-stage one. The HYPE Index starts from the other end. It identifies the strongest claim an entity makes in its most prominent public channel, then asks how far that claim sits from what the evidence supports. The gap is the score. Three consequences follow. An entity with little evidence can score as calibrated if it says little. An entity with substantial evidence can score as hype if it says more than that evidence covers. And the same words can be calibrated from one entity and hype from another, because the appropriate calibration depends on the stage.
The four axes
Each axis is named by a pair of words that encode the tension it measures.
| Axis | Question it asks | Criteria (each scored 0 to 4) |
|---|---|---|
| H: Hypothesis and History | Did this start as a real, testable question, and does it have a paper trail? | Hypothesis articulation; historical and failed analogues acknowledged; appeal to the appropriate verifier; track record transparency; distance between the hypothesis tested and the claim sold |
| Y: “Yes, but” and Yields | Yes, it works, but in whom, where, and by how much? | Population specified; setting specified; claim scope matches regulatory scope; real-world yield (the absolute effect actually demonstrated); outcome metrics versus surrogate or process metrics |
| P: Peer Review and Proof of Work | Have qualified outsiders validated this, and has the work been shown? | Peer-reviewed trials exist; dose, form and standardization disclosed; funding and data provenance disclosed; independent non-manufacturer trials; replication or third-party validation |
| E: Expectation and Evidence | What is the strongest thing promised, and what has actually been earned? | Strongest public claim identified; evidence-claim correspondence; marketing language versus methods language; reliance on influencer amplification; update behavior when contradicting evidence appears |
E is the headline axis. The other three provide the diagnostic context that explains why a gap exists. On every criterion, 0 is the strongest possible calibration and 4 is the largest possible gap, with anchored verbal descriptors so two raters scoring the same material can see exactly where they disagree.
Stage and verifier
Before anything is scored, it is classified by stage, because the appropriate calibration differs by stage. For supplement claims: stage 0 is mechanistic or preclinical only; stage 1 is pilot human data; stage 2 is randomized trials on surrogate markers such as HbA1c or LDL; stage 3 is randomized trials on clinical outcomes such as polyp recurrence or heart attacks. Each claim is also matched to its verifier, the mechanism in the world that could settle it. In descending order of strength: a clinical outcome, a regulatory clearance, peer-reviewed trials, an experimental platform, or no verifier at all. Some claims, especially longevity claims, have no verifier within a human lifetime. The Index says so rather than scoring them as if a trial could be run.
From criteria to a score
The five criteria on each axis sum to an axis score from 0 to 20. The composite weights the headline axis double:
Composite = (0.2 H + 0.2 Y + 0.2 P + 0.4 E), normalized to 0 to 100.
A composite above 60 earns the Significant Evidence Gap (SEG) label. The label is editorial commentary on the size of the gap, not a statement about anyone’s intent, and a 59 and a 61 are nearly the same thing. For narrative coverage, scores fall into five tiers:
| Composite | Tier | Reading |
|---|---|---|
| 0 to 20 | Calibrated | The claim matches or undersells the evidence |
| 21 to 40 | Largely calibrated | Minor drift, usually in scope or population |
| 41 to 60 | Mixed calibration | Real evidence, real overreach, both at once |
| 61 to 80 | Significant Evidence Gap (SEG) | The claim runs well ahead of what has been shown |
| 81 to 100 | Substantial Evidence Gap | The claim has no meaningful evidentiary support at its stage |
The chain of hype
A single finding travels. A paper reports that a compound lowered a biomarker in a specific population over twelve weeks. A label says the compound “supports healthy” something. A founder on a podcast says it “works like” a drug. A clipped caption says scientists found “nature’s” version of that drug. Each link can be scored on the same rubric, which makes the drift measurable and locates where it happened. On supplement pages, the chain-of-hype note under the HYPE table names the link where a claim left its evidence behind.
How supplement pages apply it
Every ingredient review on Healthcare Discovery is generated from a structured evidence database rather than written as prose. For each ingredient, the pipeline pulls randomized trials, systematic reviews and meta-analyses from PubMed, extracts design, sample size, duration, population, intervention, comparator, effect size and funding from each, then decomposes the ingredient’s marketing into discrete claims. Each claim receives four axis scores, a composite, a stage, a verifier, and a plain-language evidence grade (strong, moderate, preliminary, insufficient or contradicted) that summarizes the P and E axes for readers. Doses appear only as reported by named sources, never as recommendations. Interactions are compiled from the trial exclusion criteria and the pharmacology literature. New studies indexed in PubMed are added monthly and dated on each page under “New since our last review.”
What the score is not
The score is not a measure of whether an ingredient works. Berberine’s blood-sugar claim is well supported and scores as largely calibrated; its longevity claim is unsupported and scores as a substantial gap; both describe the same molecule. The score is not a verdict on a company or an author. It is not clinical guidance: Healthcare Discovery synthesizes evidence and lets readers draw conclusions with their clinicians. And it is not static. Scores change when trials are published, which is the point of building them from a database rather than a draft.
Limits and open work
Scoring from abstracts rather than full methods sections can miss limitations; flagship articles are scored from full text. Inter-rater reliability testing across raters and across independent AI systems is in progress and will be published as it completes. The methodology is a living document, versioned, with a correction log on this page.
Version and corrections
HYPE Index Methodology v1.1 (Diligence Dx working draft, May 2026), adapted for supplement claims October 20, 2026. Corrections: none yet. Questions about a specific score belong in the comments of the page that carries it.
See it applied: Berberine Review 2026: Evidence Grade, Best Form, Side Effects.
