A Hybrid Approach to Developing a Stroke Prediction System

Abstract

The development of a stroke prediction system using machine learning algorithms offers a novel approach to identifying individuals at risk for stroke. By analyzing large datasets, it is possible to identify patterns and predictors of stroke that may not be apparent to human clinicians. This system has the potential to improve early detection and treatment of stroke, leading to better patient outcomes and helping to identify at-risk individuals who may benefit from preventive measures. Although single techniques have been employed to improve the accuracy and robustness of stroke prediction models, this study performs a hybrid technique using logistic regression (LR), random forest (RF), and support vector machines (SVMs) to enhance the accuracy and robustness of the proposed model. All three algorithms performed well in terms of accuracy, with random forest achieving the highest accuracy. However, LR and SVM were more efficient regarding training time and complexity. The overall conclusion was that RF is the best-performing algorithm for this particular task, but other algorithms may be more suitable for different applications. In conclusion, developing a stroke prediction system using machine learning algorithms is a promising approach for improving stroke prediction and patient outcomes. This study shows that machine learning algorithms can effectively identify individuals at risk for stroke and may have advantages over traditional risk factors. However, more research is needed to fully understand the potential of machine learning in this field and to determine the most effective algorithms and training methods.