Brain Stroke Detection Through Advanced Machine Learning and Enhanced Algorithms
P. Yogendra Prasad, M. Ramu, Komali Anitha, Kothapalli Lalasa, Dasari Hasritha, Boggula Aarthi Reddy
Abstract
Stroke, the second leading cause of mortality globally, demands timely and accurate prediction for effective intervention. This study explores advanced machine learning techniques to enhance stroke prediction models. Initially employing Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM) classifiers, later the research introduced eXtreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LGBM) algorithms to bolster predictive capabilities. Evaluation metrics including accuracy, sensitivity, error rates, and log loss were utilized to assess model performance. Findings highlight the efficacy of machine learning algorithms, with XGBoost achieving remarkable accuracy of 98%. Complementarily, LGBM significantly contributed to overall accuracy. These results emphasize the pivotal role of advanced machine learning techniques in improving stroke prediction. Utilizing cutting-edge predictive models driven by advanced algorithms, this study advocates for their seamless integration into clinical practice. By doing so, it aims to expedite precise diagnoses, thereby enhancing patient care and pushing the boundaries of stroke detection forward.