Comparative Analysis of Various Machine Learning Approaches to Predict Postoperative Mortality in Patients Undergoing Aortic Valve Replacement for Severe Aortic Stenosis

Severe Aortic Stenosis (AS) is a life-threatening condition that progressively worsens with age and lifestyle factors, mainly affecting the elderly. In most cases, surgical intervention is required, followed by vigilant postoperative monitoring to ensure patient survival. This paper evaluates the predictive performance of eight machine learning algorithms in forecasting postoperative all-cause mortality in AS patients using two approaches: no resampling and resampling with the Synthetic Minority Oversampling Technique (SMOTE) combined with Tomek Link removal. The predictive analysis incorporates two key cardiovascular biomarkers: hsTnT and NT-proBNP. Eight state-of-the-art machine learning algorithms, including Logistic Regression, Support Vector Classifier, k-nearest Neighbors, Decision Tree, Gaussian Naive Bayes, Random Forest, AdaBoost and Gradient Boosting, were evaluated for their predictive performance. Results show that AdaBoost and Gradient Boosting trained on resampled data outperformed others, achieving an accuracy of 0.78. The findings indicate the potential application of machine learning techniques in enhancing diagnostic and prognostic abilities for cardiovascular diseases as larger and more diverse datasets become available.