Leveraging Machine Learning Techniques to Forecast Chronic Total Occlusion before Coronary Angiography

Area of Science:

• Cardiology
• Interventional Cardiology
• Machine Learning in Medicine

Background:

• Chronic total occlusion (CTO) is a complex challenge in coronary artery disease (CAD) management.
• Current risk stratification for CTO lacks personalized predictive tools.
• Predicting CTO before coronary angiography (CAG) is crucial for treatment planning.

Purpose of the Study:

• To develop and validate a precision medicine tool using machine learning to predict CTO in CAD patients.
• To identify key clinical features predictive of CTO.
• To facilitate early discernment of CTO in routine clinical practice.

Main Methods:

• Utilized data from 1473 CAD patients (1105 training, 368 testing).
• Performed univariate and multivariate logistic regression to identify independent risk factors.
• Developed and validated a CTO prediction model using a machine learning algorithm.
• Evaluated model performance using the area under the curve (AUC).

Main Results:

• A CTO prediction model was developed incorporating eight important variables: gender, neutrophil percentage, hematocrit, total cholesterol, HDL, ejection fraction, troponin I, and NT-proBNP.
• The model demonstrated good predictive performance with AUCs of 0.724 (training) and 0.719 (testing).

Conclusions:

• An accessible tool for predicting CTO in CAD patients has been successfully developed and validated.
• Further research with larger cohorts is recommended to enhance the model's predictive accuracy.
• The tool can assist clinicians in making informed decisions regarding early CTO detection.