Did SignalPET's AI “match the best radiologist”?
The reported accuracy estimates were close. We trace why a non-significant difference does not demonstrate equivalent performance.
Claim investigations
Marketing claims meet evidence review.
Investigations start with a specific claim, identify exactly what would make it true, then map the evidence that supports, limits or fails to answer it.
What a claim check will contain
Each review will state the claim, source type, evidence category, independent reproduction status and material caveats.
When review becomes testing
If published evidence cannot answer a practical question, VetAI Trust may design an independent test with disclosed baselines and limits.
Investigation standard
Precise, fair and reproducible where possible.
Define the claim
We preserve the claim wording and context before interpreting what is being promised.
Map available evidence
Vendor materials, papers, validation studies and independent sources are separated rather than blended.
Identify what is missing
A missing independent replication is reported as an evidence gap, not as proof that the claim is false.
Publish with caveats
Findings disclose methodology, uncertainty and conflicts of interest.