The primary cause of death and disability worldwide is stroke. Accurate prediction models and identification of stroke risk factors can aid in early intervention and preventive measures. In this study, an approach based on machine learning algorithms, including Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), AdaBoost (AdB), Gradient Boosting (GB), Decision Tree (DT), and a unique Artificial Neural Network (ANN) model, is proposed to predict stroke risk at four different levels. To address the issue of imbalanced dataset, Synthetic Minority Over-sampling Technique coupled with Edited Nearest Neighbors (SMOTEENN) was used. 11 features were selected for training the models out of the initial 12 features, and the data-set was divided into an 80–20 train-test ratio for model assessment. The Decision Tree (DT) model achieved an accuracy of 98.53%. Additionally, a 1000 epoch Artificial Neural Network (ANN) model with various batch sizes was created; 128 produced the best results. Stochastic Gradient Descent (SGD), Adam, and RMSprop were examined along with other optimizers, however SGD produced the highest accuracy rate of 99.21%. The results suggest that utilizing the selected features and the balanced data-set, machine learning algorithms such as the unique CNN model can accurately predict stroke risk at various levels. These models have the potential to aid in early detection of stroke risk and guide preventive interventions, thus improving patient care and outcomes. Further testing and implementation of these models in real clinical settings are warranted.