PB3774 CT-Based Deep Learning to predict cardiac dysfunction in patients with acute pulmonary embolism

In patients with acute pulmonary embolism, an AI model trained on CTPA-derived pulmonary vascular features accurately predicted abnormal cardiac biomarker results, achieving an AUC of 0.88. This imaging-based approach may support early risk stratification when biomarkers are unavailable or delayed. These findings require further validation.