Stroke is the leading cause of disability in a lot of people, resulting in many financial and social issues. Stroke can be dangerous if left untreated. Individuals may obtain appropriate therapy more rapidly if monitored, and their symptoms are detected and precisely analyzed in real time. Various algorithms and models have been developed. Foreseeing will the person have a chance to get a brain stroke. The research trains various models for reliable prediction using different parameters and ML algorithms like decision tree method, RF technique, LR approach, SVM approach, LightGBM (LGBM), GNB, and XGBoost (XGB). Random forest and XGBoost were the top-performing algorithm in this research; they have given the resultant output of almost 99%. We have used a stroke prediction dataset that is openly available on Kaggle for download. The important and essential agenda of this study is to create a stacking method with great performance verified with multiple measure values, such as AUC, preciseness, and recall, as well as F1-measure and accuracy. The study results reveal that this classification surpasses all other approaches, it has AUC of 99%, precision, F1-measure, and recall is 98%, with final accuracy 99%. This study’s models are far more accurate than those used in previous studies, indicating their increased dependability.