Skip to content
type1.science

Evidence transparency

What has been checked?

Sources, review notes and an honest account of the work still open. A recent date is useful only when you know what was reviewed.

Review depth varies across the index.

The 2026-09-16 editorial pass left scoped notes for 441 of 441 current records. Those notes describe specific checks and limitations. They do not certify every assertion or establish that every worldwide programme is included.

Find a record’s review notes →

The stronger verification register

This register was introduced after the scoped September review. It starts each field and paragraph as pending until its source comparison and independent second pass are documented. Prior notes have not been automatically marked as verified.

0 / 21,726Text/data units meeting the full register checks
21,726Units still requiring complete verification
29 / 29Discovery search streams still pending in this register

Full claim-register gate: incomplete. A unit can contain several claims. These are review-work counts, not a percentage accuracy score. 71 removed assertions are still marked pending in the register; changes to that status also need a documented reason and independent review. Passing the register would establish documented checks, not clinical certification or guaranteed worldwide completeness.

This aggregate covers the existing item/trial register. The newer comparison facts, country-access profiles and reader digest have their own visible sources and check dates; they are not counted as fully reviewed item records here.

Three dates, three different meanings

Source published
When a paper, decision or announcement appeared. A new citation alone does not mean the whole record was reviewed again.
Registry checked
When the study register was consulted for status, locations or eligibility. 220 trial records have a separate registry-check date; the latest recorded date is 2026-09-16. This does not verify historical study results or confirm a place at a local site.
Editorial review
A dated review of the record within the scope described in its notes. Read those limits alongside the evidence.

What still needs work

  • Complete claim-by-claim primary-source checks and independent review, including older high-impact results and safety claims.
  • Repeat searches for missing developments, negative findings and discontinued programmes across categories.
  • Resolve inaccessible sources and missing study details. An abstract or sponsor announcement may not answer every question.
  • Extend country-specific access checks and practical comparisons, with exact product generations and dated sources.

Learn how to interpret the evidence →

Documented corrections

Selected correction entries from the published change history. These are records of changes, not a new verification of every associated statement.

2026-09-16

Diabeloop DBLG2 →

Corrected predecessor-based evidence attribution and removed unexplained score advantages over DBLG1. Shared clinical/feature ratings now use consistent predecessor anchors; phone control retains its documented form-factor benefit. AID default weights were separately rebalanced toward everyday use and access.

2026-09-16

Loop (iOS) →

Updated the September Omnipod 5 software release and clarified open-source use and support limits.

2026-09-16

Accu-Chek SmartGuide →

Corrected the claim that no fingerstick calibration is required: the system requires two initial calibrations.

2026-09-16

Teplizumab (Tzield) →

Updated the June US label, including the boxed viral-reactivation warning, and clarified age, disease-stage and country-specific access. Added the September PROTECT per-protocol analysis without replacing the original randomized secondary-null findings.

2026-09-16

CamAPS FX →

Corrected the overlooked May 2024 FDA clearance; clearance remains distinct from US commercial availability and from Liberty’s European launch.

Read the full change history →

Help improve the index

If you spot an incorrect claim, missing development or outdated source, include the record link and the supporting evidence. Reports go through editorial checking before a correction is published.

Report an issue →