An Integrative Machine Learning Model for Predicting Early Safety Outcomes in Patients Undergoing Transcatheter...

Area of Science:

• Cardiology
• Medical Informatics
• Biomedical Engineering

Background:

• Early safety outcomes after transcatheter aortic valve implantation (TAVI) are crucial for severe aortic stenosis patient prognosis.
• Predicting adverse events aids in optimizing patient management and improving TAVI outcomes.

Purpose of the Study:

• To develop a machine learning model for predicting early safety outcomes in patients undergoing TAVI for severe aortic stenosis.
• To identify key predictors of adverse events in the early post-TAVI period.

Main Methods:

• Retrospective analysis of 224 patients with severe aortic stenosis who underwent TAVI.
• Utilized 77 clinical and biochemical variables, employing ADASYN for imbalanced data and a Random Forest model.
• SHAP values were used to interpret model predictions and variable importance.

Main Results:

• The Random Forest model identified left femoral artery diameter and aortic valve calcification volume as significant predictors.
• Smaller left femoral artery diameter and increased aortic valve calcification volume were associated with poorer early safety outcomes.
• SHAP analysis confirmed the influence of these variables on predicting adverse events.

Conclusions:

• A machine learning model effectively predicts early safety outcomes following TAVI.
• Left femoral artery diameter and aortic valve calcification volume are critical factors for risk stratification.
• Integrating these predictors into pre-procedural assessments can enhance clinical decision-making and patient care.