Published Paper Details:

A brain stroke is a medical disorder where the brain is damaged due to a burst in the blood vessels in the brain. These symptoms occur if the supply of blood or any other nutritional resources to the brain are interrupted. The studies indicate that brain stroke is the primary cause of death and disability worldwide. Timely detection of strokes is essential for effective treatment. Current stroke detection techniques include machine learning, but they struggle with complex patterns in medical imaging and need human-designed features. These approaches cause slower diagnosis and erroneous outcomes. To estimate the probability of early-stage brain strokes, the paper experiments using various algorithms, including VGG16, XGBoost, GNN, and EfficientNetB3 architectures. To determine the effectiveness of the algorithms, a reliable dataset for stroke detection was taken from Kaggle website. Several classification models, including XGBoost, VGG-16, EfficientNetB3, and GNN were used. By training the model on a large dataset and classifying using different algorithms the XGBoost has produced the highest accuracy of 97.94% than the GNN, VGG-16, and EffiientNetB3 which produced an accuracy of 73.68%, 87.26%, 75.28% accuracy. This paper showed that XGBoost performed better compared to other algorithms.