Stroke is a high-risk neurological condition caused by blockages or bleeding in the brain, leading to death or disability. This study proposes a model to address the imbalance in limited patient data.The proposed model uses the MissForest method, a Random Forest Regression algorithm, to complete missing data and an artificial immune system algorithm whose parameters are updated using the Firefly algorithm to stabilize the data. The One-Sided Selection model is used to improve the performance of the minority class.The model was evaluated in two experiments, one using all features and the other selecting the best features using the Artificial Bee Colony (ABC) algorithm. The models were trained using six different classification algorithms: CatBoost, Light Gradient Boosting Machine (LightGBMBoost), Gradient Boosting (GB), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), and Logistic Regression (LR). The results were presented using performance metrics. When trained using all features, the model achieved an accuracy of 77%, specificity of 44%, and sensitivity of 77%. When trained using the best features selected by the ABC algorithm, the model achieved an accuracy of 81%, specificity of 61%, and sensitivity of 81%. Compared to previous studies, the proposed model was effective in both experiments.