Thymosin beta-4 · Sensory · graded 2026-07-30
Eye & vision health
This grade describes how much evidence exists for Thymosin beta-4 against eye & vision health — not whether it works, and not whether it is safe. It is computed from the inputs below by a fixed rule with no editorial adjustment.
Why this grade
Not hidden behind a control — showing the work is the point| Component | Points | Scale |
|---|---|---|
| Best study type The strongest design that produced evidence for this outcome, not the average design. | +5 | Human RCT 5 · Human observational 4 · Animal 2 · Preclinical 1 · None 0 |
| Largest human sample Largest single human study, not a pooled total across studies. | +1 | n ≥ 200 → 3 · 50–199 → 2 · < 50 → 1 · no human study → 0 |
| Independent replication Distinct research groups at the best study tier above, resolved through author and institution records. | +1 | ≥ 3 groups → 3 · 2 → 2 · 1 → 1 · 0 → 0 |
| Directional consistency Share of studies at the best study tier reporting effect in the same direction — the same studies the replication row counts, not every record on the pair. Measured over 0 stated directions across 1 study at the best tier. A ratio over three readings and the same ratio over ninety are the same number and are not the same evidence. | 0 | ≥ 90% concordant → 3 · ≥ 70% → 2 · ≥ 50% → 1 · below → 0 · fewer than 3 readings → 0 |
| Bias risk A deduction, never a bonus. A flag is counted only where a record states one; silence is not a pass and is not a flag. | -1 | −1 per flag, floor −3: industry funding · no preregistration · no blinding |
| Total | 6 | Sum of the rows above, floored at 0. |
| Field | Flag stated | Stated the opposite | Not stated | Silent |
|---|---|---|---|---|
| industry_funded industry-funded deducts a point; declared non-commercial does not; silence does neither. | 0 | 1 | 17 | 94% |
| preregistered not preregistered deducts a point; preregistered does not; silence does neither. | 1 | 0 | 17 | 94% |
| blinded not blinded deducts a point; blinded does not; silence does neither. | 0 | 1 | 17 | 94% |
Grade A additionally requires a total of at least 11 and replication of at least 2: a high score carried by one research group is not replicated evidence, however large the trial.
Doses administered in studies
Sourced figures, not a protocolStudied doses — not yet collected
Dose, route and n are extracted from study full text, and full text is paywalled for most records in this corpus. No dose has been extracted for this pair yet. Read that as an extraction gap, not as evidence that the studies below administered nothing.
This is a gap in our collection, not a finding about studied doses. Read it as “we have not looked yet”, never as “there is nothing there”. How extraction works and where it fails
Cited sources
18 records · ordered by source hierarchyThese are the records the grade above was computed from — the same set, by the same rule, read from one definition shared with the grader. Records attributed to an alias for a materially different product are not among them; where this pair has any, they are listed separately below.
- HUMAN-TRIAL1
- ANIMAL5
- PRECLINICAL12
Rows are ordered by source hierarchy, so nothing below a regulatory filing appears above a human trial. A field marked uncertain is one where the extraction passes disagreed; it publishes as unknown rather than as a picked winner. A design reading “not determined” is a design nothing established, not a design of none. Explore the full corpus for Thymosin beta-4