New data suggest AI, specifically metabolomics-based machine learning models, could revolutionize screening for metabolic dysfunction-associated steatohepatitis (MASH) candidates for resmetirom. Current noninvasive tests for noncirrhotic MASH with moderate-to-severe fibrosis have limitations, but machine learning models show potential to improve accuracy, sensitivity, and predictive ability. The study, led by Christos Mantzoros, assessed 28 biomarker-, imaging-, and algorithm-based noninvasive tests and developed machine learning models achieving higher AUC, sensitivity, and NPV. The addition of specific biomarkers to metabolomics-based models further enhanced performance.