ResearchBenchmarksCase-grounded verification on MIMIC-CXR (test)Follow97.43AUROCS358.398868.531978.66588.7981Apr 10, 2026Evaluation ResultsMethodMethodLinksAUROCAUPRCF1 ScoreAccuracyBalanced AccuracySensitivitySpecificityPrecisionBrier ScoreS3Input=Case + Evidence...Input=Case + Evidence + Claim2026.0497.4387.7878.7291.2292.0393.2890.7968.180.06S2Input=Evidence + ClaimInput=Evidence + Claim2026.0482.6335.1162.7979.487.5310075.0745.780.255S1Input=Case + ClaimInput=Case + Claim2026.0459.926.2332.2165.7458.447.1469.6625.460.201