A Review - On Brain Stroke Prediction

Abstract

Recognizing and treating strokes early is crucial for improving outcomes, which is why stroke remains a significant public health issue. Over the years, numerous studies have aimed to create dependable methods for detecting brain strokes, especially through the use of machine learning techniques. Although, these initial efforts frequently struggled because of the small size of the datasamples available. The research review takes a closer look at the less amount of datasets used in the early detection of brain stroke disease, emphasizing how datasamples size affects performance. This study examines various observations that utilized different types of datasamples, which contain medical information and imaging data, and combinations of medical information and imaging data. The observation of this study reveal that the accuracy of brain stroke disease detection was notably hindered by the small dataset sizes in earlier research.  Over fitting and the models poor generalizability plagued some studies, despite their promising outcomes. These studies demonstrate the accuracy and dependability of the models related to brain stroke detection have increased significantly. This paper focuses on the importance of the proportion of a dataset when constructing dependable models for brain stroke detection.