Enhanced Machine Learning Algorithm to Predict Brain Stroke

A stroke is a condition when there is insufficient blood flow to the brain, which ultimately leads to cell death. It is nowadays a leading cause of dying all over the world. An analysis of those experiencing strokes determined a multitude of medical conditions that have been designated as possibly associated with the root cause of the stroke. Although many individuals have died from strokes, as much as 85 percent of strokes can be avoided by discovering or detecting them in their early stages. Leveraging these risk factors, numerous research studies have been conducted to predict the conditions linked to stroke. The majority of these predictive models are constructed using machine learning techniques. In previous research, they have trained the models for accurate prediction using machine learning methods including Logistic Regression Algorithm, Decision Tree Classification Algorithm, Random Forest Classification Algorithm, K-Nearest Neighbors Algorithm, Support Vector Machine Algorithm, and Naive Bayes Classification Algorithm. A textual (.csv) training dataset was used for this work. The algorithm that performed this task most accurately was Support Vector Machine, which achieved an accuracy of almost 94.6%. In this proposed analysis, the stroke prediction process has been advanced by employing Random Forest Classification and XGBoost Algorithm methods. These approaches are augmented with feature selection techniques like Chi-Square and Information Gain to elevate the accuracy of the predictive models. The textual dataset is used in this study. Both Random Forest Classification and XGBoost Algorithm, which achieved an accuracy of about 98%, was the algorithm that performed these tasks best.