STEMI

STEMI predicts intrahospital mortality risk in patients with ST-segment elevation myocardial infarction complicated by hyperuricemia using nomogram-based modeling with LASSO variable selection and complementary machine learning algorithms.

Key Features:

• Predictive Modeling: Nomogram-based predictive model developed using the least absolute shrinkage and selection operator (LASSO) to identify significant predictors from demographic and clinical variables.
• Variable Selection: Incorporates biomarkers and clinical variables including B-type natriuretic peptides, α-hydroxybutyrate dehydrogenase (α-HBDH), cystatin C, out-of-hospital cardiac arrest (OHCA), shock index, and neutrophil-to-lymphocyte ratio.
• Model Performance: Demonstrates discrimination with area under the curve (AUC) of 0.875 in the training set and 0.87 in the validation set and was evaluated with calibration plots and decision curve analysis.
• Machine Learning Integration: Implements machine learning algorithms that outperform the Thrombolysis in Myocardial Infarction (TIMI) score and identify distinct factors relevant to short- and long-term mortality in multi-ethnic populations.
• Population-Specific Predictions: Analysis of a heterogeneous Asian population highlights the importance of invasive management variables for improving mortality predictions.
• Continuous Validation: Includes ongoing testing and validation to refine risk stratification across different datasets.

Scientific Applications:

• Personalized Therapy Management: Provides individualized intrahospital mortality risk assessments to inform therapeutic decisions for STEMI patients with hyperuricemia.
• Enhanced Risk Stratification: Identifies specific predictors of mortality to improve patient classification and management, particularly in diverse populations.
• Clinical Decision Support: Integrates complex clinical and biomarker data into actionable risk estimates to support clinician decision-making.

Methodology:

LASSO-based variable selection was used to construct a nomogram; model evaluation employed AUC metrics, calibration plots, and decision curve analysis; additional machine learning algorithms were developed and compared against the TIMI score.