Hybrid Features Binary Classification of Imbalance Stroke Patients Using Different Machine Learning Algorithms

Abstract

Nowadays the quantity of paralyzed patients has been increasing due to nervous impairments, spinal cord injuries and stroke. These types of patients required friend and family support for rehabilitation to enhance their lives. All clinicians are highly desirable to predict post-stroke functional outcomes. Analyzed stroke patients' predictions based on gender, income rate, private and public job including heart and diabetic diseases. Synthetic Minority Over-Sampling Technique (SMOTE) is used on our Imbalanced data and compared with Over-Sampling and Down-Sampling by using different Machine Learning Algorithms to predict stroke. After comparing XGB-Classifier with 84% accuracy is best on Unbalanced data, almost 99% accuracy shows on Random forest classifier and XGB-classifier on over-sampling, lastly in down-sampling almost all algorithms give 100% accuracy.