<p>Hypertension (HTN), despite contemporary endovascular repair, is a common and challenging complication of coarctation of the aorta (CoA), and its mechanisms and optimal management remain uncertain. Using computed tomography angiography (CTA), we present a feasibility workflow that integrates statistical shape analysis (SSA), computational hemodynamics, and machine learning (ML) to investigate predictors of HTN persistence after endovascular treatment. It builds on our randomized controlled trial comparing safety and efficacy of two types of aortic stents, in which all patients underwent a 3-year structural follow-up with blood pressure measurements, transthoracic echocardiography, and CTA. The current analysis includes twenty-nine patients with paired baseline and follow-up CTAs. Deep-learning segmentation was used to reconstruct patient-specific aortic geometries, from which statistical shape modes (SSMs) were derived. In addition, CFD-based hemodynamic indices were computed to characterize simulated flow patterns. These features were then evaluated using a stacking ensemble classifier and complementary nonparametric statistical testing to predict HTN at 3-year post-procedure. In four-fold cross-validation, model performance varied across folds, with accuracies ranging from 71.9 to 93.8% and area under the receiver-operating-characteristic curve (AUC-ROC) ranging from 0.74 to 0.95. Statistical analysis also identified several hemodynamic variables as candidate biomarkers associated with post-treatment HTN persistence. Overall, these results support the feasibility of combining SSA, computational hemodynamics, and ML to explore shape- and flow-related factors associated with post-repair HTN.</p>