Predictive Analysis for Risk of Stroke Using Machine Learning Techniques

Stroke occurs when our brain's blood flow is stopped or reduced, restricting brain tissue from receiving oxygen and important nutrients. Treatment of stroke disease is very crucial. So, the prediction of stroke is significant for early intervention and treatment. Stroke can be predicted by analyzing different warning signs. In this experiment, we implement a process of stroke risk prediction from our dataset using the various machine learning algorithms. Data imputation, feature selection, data preprocessing is considered as the initial job in this work. Some physical parameters such as heart disease, age, BMI, gender, hypertension, etc. are considered as the feature is used to model training and testing. In this study, adaBoost classifier, artificial neural network, decision tree classifier, k-nearest neighbour (KNN) classifier, random forest, stochastic gradient descent (SDG), support vector machine (SVM), XGBoost classifier are implemented to predict the risk possibility of stroke. Then voting classifier is implemented on these eight traditional classifiers. After analyzing different machine learning algorithms with voting classifiers, we found 98% accuracy to predict the risk factor of stroke, which is better than other models.