Predictive value of CAC score combined with clinical features for obstructive coronary heart disease on coronary...

Area of Science:

• Cardiology
• Artificial Intelligence in Medicine
• Medical Imaging Analysis

Background:

• Obstructive coronary heart disease (CAD) diagnosis often relies on invasive procedures.
• Atypical chest pain presents a diagnostic challenge, necessitating accurate non-invasive risk assessment.
• Coronary computed tomography angiography (CCTA) and coronary artery calcium (CAC) scoring are established non-invasive tools.

Purpose of the Study:

• To evaluate the predictive performance of a machine learning (ML) model integrating clinical factors and CAC score for obstructive CAD.
• To compare the ML model's accuracy against traditional logistic regression (LR) models using CCTA data.
• To identify key predictors for obstructive CAD within the ML framework.

Main Methods:

• A cohort of 1,906 patients with atypical chest pain and no prior CAD underwent CCTA and CAC scoring.
• A Random Forests (RF) model was developed using 63 variables, including clinical factors, CAC score, and laboratory/imaging parameters.
• The RF model was trained on 70% of the data and validated on the remaining 30%, with performance compared to two LR models.

Main Results:

• The incidence of obstructive CAD was 16.4%.
• The RF model achieved a superior Area Under the Receiver Operator Characteristic curve (0.841) compared to the CAC score model (0.746) and clinical model (0.810).
• Key predictors identified by the RF model included CAC score, age, glucose, homocysteine, and neutrophil count.

Conclusions:

• The Random Forests ML model demonstrates superior predictive capability for obstructive CAD compared to traditional logistic regression.
• This ML approach offers potential for improved risk stratification and personalized management of patients with suspected coronary artery disease.
• Integration of CAC score and clinical data via ML can enhance diagnostic accuracy in challenging patient populations.