LASSO Regression-Derived Nomogram of Cognitive Frailty Following Ischemic Stroke Based on Self-Perceptions of Aging

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background: Cognitive frailty (CF) is a high-risk state associated with poor long-term outcomes in elderly ischemic stroke (IS) patients. Objectives: The purpose of this study was to develop and validate a nomogram incorporating self-perceptions of aging (SPA) and clinical variables to predict individualized risk of CF in elderly IS survivors. Methods: This study included 940 elderly IS patients (≥60 years) who were randomly split into a training set (70%, n = 658) and a validation set (30%, n = 282). The optimal predictors were identified by taking the intersection of variables selected through Least Absolute Shrinkage and Selection Operator (LASSO) regression, univariable logistic regression, and multivariable logistic regression, and a nomogram was subsequently developed to visualize the model. Model performance was evaluated using area under the curve (AUC), calibration plots, and decision curve analysis (DCA), with validation by a gradient boosting machine (GBM) model. Results: Eight predictors (age, living with family members, family history of cognitive decline, physical activity, previous stroke ≥2 times, SPA, cancer, chronic kidney disease) were retained. The nomogram achieved AUCs of 0.913 in the training set and 0.899 in the validation set, with excellent calibration. Calibration curve and DCA demonstrated clinical utility across threshold probabilities. GBM confirmed SPA as the most influential predictor. Discussion: This nomogram, integrating SPA and clinical factors, provides a robust tool for predicting CF in elderly IS patients, supporting early intervention and personalized care strategies.