A prediction model for the risk of gastrointestinal bleeding associated with antiplatelet therapy in patients with ischemic stroke

Objective: Patients with ischemic stroke (IS) undergoing antiplatelet therapy are at risk of gastrointestinal bleeding (GIB). This study aims to develop and validate a multivariable integrated risk prediction model for GIB, to optimize clinical decision-making. Methods: A retrospective cohort of IS patients who received antiplatelet therapy from 2020 to 2024 was included. Demographic characteristics and laboratory parameters (including complete blood count, coagulation profile, liver and kidney function tests, and stool occult blood) were collected. Predictive factors were selected using LASSO regression and logistic regression, and a nomogram model was constructed. Evaluation metrics included area under the curve (AUC), calibration curve (mean absolute error, MAE), and decision curve analysis (DCA). Results: Six independent risk factors were identified: C-reactive protein (CRP) (p = 0.003), hemoglobin (HGB) (p < 0.001), D-Dimer (p = 0.039), albumin/globulin ratio (ALB/GLB, p = 0.021), age (p = 0.01), and fibrinogen (FIB, p = 0.037), which collectively drive the risk of GIB. The predictive model demonstrated an AUC of 0.79 in both the training and validation cohorts, with MAE values ranging from 0.018 to 0.04, and a Hosmer-Lemeshow test result of p > 0.05. The model exhibited good fit, strong discrimination capability for GIB, and stable diagnostic performance. Decision curve analysis revealed significant net benefits within the risk threshold range of 0.2-1. Conclusion: The developed nomogram model effectively predicts the risk of GIB in IS patients undergoing antiplatelet therapy, providing a basis for individualized treatment strategies.