A Machine Learning Model Using Cardiac CT and MRI Data Predicts Cardiovascular Events in Obstructive Coronary Artery...

Area of Science:

• Cardiovascular Imaging
• Machine Learning in Medicine
• Prognostic Biomarkers

Background:

• Multimodality imaging is crucial for assessing coronary artery disease (CAD) risk.
• Machine learning (ML) can integrate complex data for improved prognostic stratification.
• Current methods for predicting major adverse cardiovascular events (MACE) in newly diagnosed CAD require enhancement.

Purpose of the Study:

• To evaluate an ML model combining coronary CT angiography (CCTA) and stress cardiac MRI for MACE prediction.
• To compare the ML model's performance against established risk scores and individual imaging modalities.
• To validate the ML model's predictive capability on independent patient cohorts.

Main Methods:

• Retrospective analysis of symptomatic patients with newly diagnosed CAD undergoing CCTA and stress cardiac MRI.
• Development of an ML model using clinical, electrocardiogram, CCTA, and cardiac MRI parameters.
• Automated feature selection (LASSO) and XGBoost algorithm for model building.
• External validation on two independent datasets; performance assessed by Area Under the Curve (AUC).

Main Results:

• The ML model achieved an AUC of 0.86 for MACE prediction in the primary cohort.
• The ML model significantly outperformed existing scores (ESC, QRISK3, Framingham) and individual imaging data (CCTA, stress MRI).
• The model demonstrated robust performance in external validation datasets (AUCs of 0.84 and 0.92).

Conclusions:

• An ML model integrating CCTA and stress cardiac MRI data offers superior MACE prediction in newly diagnosed CAD.
• This multimodality ML approach provides a more accurate prognostic stratification than traditional methods.
• The findings support the clinical utility of advanced ML models in personalized cardiovascular risk assessment.