<p>Ischemic stroke, caused by a blockage in cerebral blood vessels, is a leading cause of mortality and long-term disability worldwide. Early prediction of stroke risk is crucial to improve clinical outcomes and enable timely interventions. Traditional prediction methods often face limitations such as low accuracy, data redundancy, and an inability to capture complex feature interactions. To address these challenges, this work presents a novel framework, the Steerable Graph Neural Network with Skill Optimization Algorithm (SGNN-SkillOA), for predicting ischemic stroke outcomes. Initially, Clinical and demographic data is collected from the Stroke Prediction Dataset. Raw data undergo preprocessing using Adjusted Min‐Max with Decimal Scaling and Statistical Column Normalization (AMM-DS-SCN), which normalizes, scales, and reduces noise for robust feature representation. Key features are then selected via the Wolf-Bird Optimizer (WBO), improving model efficiency and reducing redundancy. The preprocessed and selected features are fed into the Steerable Graph Neural Network (SGNN), which models complex feature interdependencies to predict stroke risk. Model parameters are further optimized using the Skill Optimization Algorithm (SkillOA), ensuring enhanced performance and convergence. The performance of the proposed framework is measured on stratified 80:20 train-test split and fivefold cross-validation to make the performance estimation robust. The updated findings illustrate competitive performance, better generalization to improve balancing accuracy, precision, recall and F1-score. The suggested SGNN-SkillOA model offers a valid and understandable model of early stroke prediction and clinical decision-making.</p>