Abstract
INTRODUCTION: Post-stroke cognitive impairment/dementia are disabling outcomes, yet interpretable prognostic tools remain limited. We aim to develop prediction tools accounting for sex-specific risk profiles. METHODS: We analyzed 766 stroke patients (mean age, 79.1 years; 377 men, 389 women). Shapley Variable Importance Cloud (ShapleyVIC) identified stable predictors, which were fed to the AutoScore framework to construct the Monash Stroke Dementia Score (MSDS). Model performance was evaluated using an area under the receiver operating characteristic curve (AUC), along with sensitivity, specificity, positive predictive value, and negative predictive value. RESULTS: The MSDS achieved an AUC of 0.81 (95% confidence interval [CI], 0.78-0.84), with a sensitivity of 0.78 and specificity of 0.77. Although the overall risk of post-stroke dementia was similar between men and women, sex-specific models demonstrated improved discrimination and distinct risk profiles. DISCUSSION: The MSDS provides a robust, interpretable tool for individualized prediction of post-stroke cognitive impairment/dementia, with distinct sex-specific risk patterns.