Applying Machine Learning Technique on Registry Data to Assess Factors Associated with In-Hospital Mortality Among STEMI Patients with Hypertension

This study aimed to identify factors associated with in-hospital mortality among 8,005 STEMI patients with hypertension using the ASPECT registry data. The logistic regression (LR) approach identified 12 factors, including age, male gender, diabetes, and cardiogenic shock at PCI, while the classification and regression tree (CART) approach identified 4 factors: age, diabetes, eGFR, and cardiogenic shock at PCI. Both models showed similar performance, with a Gini-index of 0.73 and an area under the precision recall curve of 0.13. Common factors identified by both methods were age, diabetes, and cardiogenic shock at PCI, highlighting the need for collaboration between biostatisticians and clinicians to better understand these associations.