Abstract
BACKGROUND: Ischemic stroke is a common and severe complication in patients with coronary artery disease (CAD), yet its early prediction remains challenging. This study aimed to investigate the association between the Fibrinogen-D-dimer to Albumin-Platelet Ratio (FDAPR) and the risk of ischemic stroke in CAD patients, and to develop a multilayer perceptron (MLP)-based machine learning model for stroke risk prediction. METHODS: Patients with confirmed CAD who underwent their first coronary angiography in two hospitals were retrospectively enrolled. The Boruta feature selection algorithm and multivariable logistic regression were used to identify independent predictors of ischemic stroke, which were subsequently incorporated into the MLP model. Model performance was evaluated using receiver operating characteristic (ROC) curves, area under the curve (AUC), and calibration plots. The Shapley Additive Explanations (SHAP) method was applied to quantify the contribution of each variable to model output, and two-dimensional contour plots were generated to explore the interaction effects between FDAPR and other key variables. RESULTS: A total of 1844 CAD patients were included in the analysis. Compared with the non-stroke group, patients with ischemic stroke were older, had higher FDAPR levels, and lower serum potassium concentrations. Boruta and multivariable logistic regression identified age, FDAPR, serum potassium (K), and total bilirubin (TBIL) as independent risk factors for stroke. The MLP model achieved an AUC of 0.71 in both the training and test datasets. SHAP analysis demonstrated that elevated FDAPR levels were positively associated with stroke risk. Contour plots further revealed synergistic interactions between FDAPR and age, potassium, and TBIL. CONCLUSION: FDAPR is significantly associated with ischemic stroke risk among patients with coronary artery disease.