AI-measured myocardial fibrosis on cardiac MRI for prognosis in cardiomyopathy: sparse evidence base

Across only two studies, AI-derived myocardial fibrosis on cardiac MRI predicted adverse cardiovascular events (ischemic HR 2.35, 95% CI 1.33-4.15; nonischemic HR 2.00, 95% CI 1.32-3.03), but certainty was very low and added value over manual quantification remains unproven.

• Two included studies: one ischemic cardiomyopathy registry (n=216) and one nonischemic dilated cardiomyopathy preprint (n=347).
• AI-derived dense late gadolinium enhancement scar improved arrhythmic event prediction beyond guideline criteria (AUC 0.63 to 0.68; p=0.02), but only in univariable analysis.
• Both studies were at high risk of bias with data-derived thresholds and no external validation; GRADE certainty was very low.

Evidence grade: very low

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.