Early Prediction of Brain Stroke Using Machine Learning Techniques

Abstract

Brain stroke is the second-leading cause of death and the third-leading cause of disability worldwide. A stroke occurs when the blood circulation in the brain is obstructed or when a blood vessel in the brain ruptures and leaks. A stroke is a medical emergency that must be treated as soon as possible. Early intervention can help to prevent brain damage and other complications. Machine learning and data science play an important role in medical science. Using technology, we can predict a disease based on the symptoms of the human body. In this paper, we propose an intelligent system that can predict potential brain strokes with only twenty-three (23) features. In addition, we apply six (10) well-known machine learning algorithms to Bangladeshi datasets collected from various hospitals in Bangladesh to assess prediction accuracy. In our work, the accuracy of gradient boosting classification is 96.09%, and it is consistent. Gradient Boosting's accuracy is higher than other classifiers such as Random Forest, Bagging, Logistic Regression, SVM, K Neighbors, Decision Tree, Gaussian Naïve Bayes, XG Boost, and Ada Boost. We have a data shortage because we only collected data from 385 patients. We can get a better result if we can manage more patients' data.