Brain Stroke Prediction using Decision Tree Algorithm

Stroke is a condition that occurs due to interruption or reduction in the blood supply to the brain. Due to this, the neurons in the brain suffer hypoxia which leads to cell injury or even death of the brain cells if not treated in time. This causes permanent or long-term injury and affects the person’s life forever. Hence, early detection and prevention of stroke are essential as it is one amongst of the top causes of mortality and morbidity worldwide. Decision tree methods have been developed as a useful tool for this and machine learning (ML) models have shown promise in stroke prediction. Decision trees remain suitable for use in healthcare applications because they offer interpretability and can handle both categorical and numerical data. A dataset containing demographic figures, medical record, lifestyle features, and other pertinent variables are used by this ML model. The data set is analyzed using decision tree method, which identifies risk factors for strokes and captures intricate interactions between predictors. The handling of missing values and outliers eliminates the need of intensive preparation. By using different parameters in the decision tree algorithm this model has successfully given about 93.4% accuracy on the validation dataset. This accuracy was given when the depth of the tree was set to 4. Depths of 9 and 16 were also implemented. The depth of 9 gave us 91.4% accuracy on the validation data whereas the depth of 16 gave us 91% accuracy. It is to conclude that adopting a simple model by applying the decision tree approach provides us the best results after training this model on various parameters.