Machine-Learning Application for Detection and Prediction of Stroke Risk and Post-Stroke Outcomes: A Comprehensive Review

Abstract

The integration of digital technologies in healthcare has piloted in a new era of utilising machine learning models, enabling early detection and prediction of disease occurrences. Despite these advancements, efficiently predicting and detecting the risk of stroke and post-stroke outcomes remains a significant challenge. Stroke often goes undetected until after it has occurred, and its potential consequences are largely unknown. Researchers have proposed using Machine-Learning (ML) models to detect and predict stroke occurrence and post-stroke outcomes to address these challenges. To the best of our knowledge, while there has been considerable research on stroke prediction, a thorough combination of ML applications intended for both stroke prediction and post-stroke outcomes remains notably absent in the literature. This paper offers an extensive literature review of ML applications in stroke prediction, post-stroke outcomes, contributory risk factors, and detection and prediction methodologies, aiming to bridge existing gaps in knowledge within this domain. This study assessed forty-six research studied authored from 2016 to 2024, aiming on ML applications for detecting and predicting stroke risks and/or post-stroke outcomes, carefully chosen from broad range of over 800 research studies through a comprehensive analysis of both quantitative and qualitative data. By synthesising and analysing research outcomes of these studies, this study intent to deliver perspective on the present status of research in this domain, identify key trends and challenges, and offer recommendations for future research directions. This review paper aims to enrich the current body of knowledge in the domain of stroke prediction and post-stroke outcomes through the lens of ML applications. Its overarching goal is to enhance early detection and predictive methodologies, thereby lessening the impact of stroke in Africa and globally.