Catalyst Grant application: the Evidence Kernel test

I checked four statements against live OpenAlex metadata, and all four outcomes matched the expected results:</p><ul><li><b>Rejection:</b> the claim relies solely on Wakefield’s 1998 paper, which has been retracted.</li><li><b>Rejection:</b> “hydroxychloroquine is effective” — its only support, Gautret 2020, has been retracted, while RECOVERY and WHO Solidarity contradict it.</li><li><b>Delay:</b> “two independent cohort studies” (Madsen 2002 and Hviid 2019) — the articles share authors, so independence is not established.</li><li><b>Pass (control case):</b> RECOVERY and Solidarity — two independent lines of evidence.</li></ul><p dir="ltr">Outcome metrics: analyst sign-off time; false-assertion rate; escalation rate; share of conclusions traceable to evidence and decision rule</p>