Prediction of All-Cause Mortality Following Percutaneous Coronary Intervention in Bifurcation Lesions Using Machine...

Area of Science:

• Cardiovascular Medicine
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Prognostic stratification after coronary bifurcation percutaneous coronary intervention (PCI) remains a significant clinical challenge.
• Machine learning (ML) offers potential for refining outcome predictions in complex cardiovascular procedures.

Purpose of the Study:

• To develop and validate a machine learning-based risk stratification model for predicting all-cause mortality.
• The model aims to integrate clinical, anatomical, and procedural features following contemporary bifurcation PCI.

Main Methods:

• Development of multiple ML models using a training cohort (n=1795) from the RAIN registry.
• Selection of 25 patient/lesion features for model training.
• Validation of the best model in an internal cohort (n=598) and external cohorts (n=1701) from DUTCH PEERS and BIO-RESORT trials.

Main Results:

• The ML model achieved an AUC of 0.79 for predicting 2-year mortality in the overall population.
• Internal and external validation AUCs were 0.74 and 0.71, respectively, demonstrating generalizability.
• Model performance was confirmed through risk ranking, cross-validation, and continual learning analyses.

Conclusions:

• The RAIN-ML prediction model is the first tool to combine diverse features for predicting all-cause mortality post-bifurcation PCI.
• The model demonstrates reliable performance and generalizability, offering a valuable tool for clinical risk assessment.