Machine Learning Approach on Predictive Model Establishment for In-Hospital Mortality in Acute Myocardial Infarction
Wenqiang Li1,2, Peng Lei1,3, Rongyan Dong4
1The First School of Clinical Medical, Lanzhou University, 730000 Lanzhou, Gansu, China.
Insights
Machine learning accurately predicts in-hospital mortality after acute myocardial infarction (AMI) using advanced data balancing and feature selection. This approach enhances risk assessment for patients undergoing percutaneous coronary intervention (PCI).
Area of Science:
- Cardiovascular Medicine
- Machine Learning in Healthcare
- Biostatistics
Background:
- Acute myocardial infarction (AMI) is a major global health concern, with percutaneous coronary intervention (PCI) reducing in-hospital mortality (IHM).
- Class imbalance in patient data poses challenges for accurate IHM prediction models post-PCI.
- Existing machine learning (ML) models often lack both high accuracy and personalized risk assessment capabilities for IHM.
Purpose of the Study:
- To develop and evaluate machine learning models for predicting in-hospital mortality (IHM) in patients with acute myocardial infarction (AMI) undergoing percutaneous coronary intervention (PCI).
- To address the challenge of class imbalance in predicting IHM.
- To identify robust data processing and feature selection methods for enhancing predictive model performance.
Main Methods:
- Retrospective study of 1693 AMI patients from January 2019 to December 2020.
- Data processing involved synthetic minority over-sampling technique (SMOTE) for class balancing, Boruta for feature selection, and grid search cross-validation (GSCV) for hyperparameter tuning.
- Six machine learning algorithms were implemented and evaluated using metrics including accuracy, sensitivity, precision, F1-score, AUROC, and AUPRC.
Main Results:
- A cohort of 1693 AMI patients revealed an IHM rate of 2.0% (34 patients) post-PCI.
- SMOTE, Boruta, and GSCV identified 32 independent risk factors and balanced the dataset.
- Ensemble ML algorithms, particularly Light Gradient-Boosting Machine (LightGBM), showed superior performance with an AUROC of 0.93 and AUPRC of 0.62, achieving 0.988 accuracy.
Conclusions:
- The combination of SMOTE, Boruta, GSCV, and LightGBM provides a robust framework for accurately predicting IHM in AMI patients post-PCI.
- Effective data balancing and feature selection are crucial for developing high-performing predictive models in imbalanced datasets.
- This approach offers potential for improved, personalized risk assessment in cardiovascular care.
Background:
Acute myocardial infarction (AMI) remains a leading cause of mortality and disability globally. Although percutaneous coronary intervention (PCI) has significantly reduced in-hospital mortality (IHM), the resultant class imbalance complicates accurate risk prediction. While machine learning (ML) demonstrates potential in predicting IHM, there is a lack of models that provide both high accuracy and personalized risk assessment.
Methods:
This retrospective study was conducted at the First Hospital of Lanzhou University from January 1, 2019, to December 31, 2020. We employed three data processing methods: synthetic minority over-sampling technique (SMOTE), Boruta, and grid search cross-validation (GSCV). Subsequently, six ML algorithms were implemented. Model performance was evaluated using accuracy, sensitivity, precision, F1-score, area under the receiver operating characteristic curve (AUROC), and area under the precision-recall curve (AUPRC).
Results:
The study cohort consisted of 1693 patients diagnosed with AMI, of whom 34 (2.0%) experienced IHM following PCI. After employing SMOTE to balance the dataset, 32 independent risk factors were identified using the Boruta feature selection method. Among the evaluated ML models, ensemble algorithms demonstrated superior performance. For instance, the Light Gradient-Boosting Machine (LightGBM) framework achieved a predictive accuracy with an AUROC of 0.93 (95% confidence interval (CI): 0.82-1.00) and an AUPRC of 0.62 (95% CI: 0.17-0.96). Additional performance metrics included an accuracy of 0.988, a precision of 0.625, a sensitivity of 0.625, a specificity of 0.994, and an F1-score of 0.625.
Conclusion:
Utilizing SMOTE for class balancing, Boruta for feature selection, GSCV for optimal hyperparameter tuning, and LightGBM for model development achieved strong predictive performance for IHM following AMI. These findings underscore the significance of robust processing and careful algorithm selection.
