XGBoost Model Using CMR Features Predicts Long-Term Outcomes in STEMI Patients

An XGBoost model combining CMR and clinical variables predicted long-term MACCEs in 483 STEMI patients, with microvascular obstruction as top predictor. Exact performance metrics not reported; external validation lacking.

• Prospective cohort of 483 STEMI patients (98 events) with median 89-month follow-up; models built with 24 demographic, clinical, and CMR features.
• Remote myocardium T1 inversely correlated with LVEF recovery at 1 month (R = -0.34, 95% CI -0.43 to -0.27, P < 0.01).
• Model superiority over other classifiers stated but without reported C-statistic or AUC; no external validation reported (likely single-center).

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