Built a stroke prediction model in R using logistic regression. Handled imbalanced data with ROSE and selected the final model based on its high sensitivity (80.6%) for clinical relevance.