Application of Support Vector Machine Model for Prediction of Stroke Vulnerability Status

Stroke is a chronic disease caused by lack of blood flow into some brain cells causing them to die due to oxygen deficiency. Cerebrovascular accidents (stroke) are the second leading cause of death and the third leading cause of disability and equally causes dementia and depression among the affected persons as well as their care takers. This disease affects people mostly at the peak of their life productive stage hence an urgent need for proactive measure through the prediction of stroke vulnerability using machine learning technique and subsequent stroke prevention. This paper aims at developing support vector machine model for the prediction of stroke vulnerability using healthcare_dataset_stroke_data obtained from Kaggle machine learning dataset repository after appropriate data preprocessing. It adequately employed the basic principles of machine learning to train the SVM model on the preprocessed dataset using python programming language. The SVM model was evaluated using python programming language sklearn evaluation metrics and the result obtained shows that support vector machine can adequately classified patients as either vulnerable or not vulnerable to stroke using the stroke risk factors profile in the dataset as evident in its accuracy and area under the receiver operating characteristics curve (AUC) of 87% and 94% respectively.