Early prediction of stroke allows for proactive measures to prevent its occurrence. This may involve lifestyle changes like eating a healthier diet, boosting physical activity, reducing stress, and quitting smoking. This paper presents a comparative comparison of predicting strokes for people based on artificial intelligence algorithms such as logistics regression, support vector machines, and random forest for a realistic dataset collected from the E Central Hospital and Soc Son General Hospital, Hanoi, Vietnam. The result provides the accuracy of LR at 0.951%, SVM at 0.955%, and RF at 0.974%. Therefore, the RF is selected to build an application for hospital centers as more tools supporting predicting stroke.