A brain stroke detection model using soft voting based ensemble machine learning classifier

Abstract

A brain stroke is a medical emergency that occurs when the blood supply to a part of the brain is disturbed or reduced, which causes the brain cells in that area to die. The process involves training a machine learning model on a large labelled dataset to recognize patterns and anomalies associated with strokes. The proposed model is an ensemble machine learning algorithm, which integrates predictions obtained from several individual classifiers like Random Forest, Extremely Randomized Trees and Histogram-Based Gradient Boosting to make a final prediction. Each classifier provides a probability estimate for each class, final prediction is based on weighted average of these probabilities. The weights assigned to each classifier can be based on their performance on a validation set or can be set uniformly. The proposed soft voting model improved accuracy and robustness of final prediction compared to a single classifier. The limitation of study classification of stroke type can lead to appropriate use of resources and help to reduce healthcare costs. To solve this issue, future introduced a swarm intelligence-based optimization for improve classification accuracy. The proposed model obtained an accuracy of 96.88%. In order to carry out the investigation, the stroke prediction dataset is collected from UCI machine learning repository.