This review discusses global trends in stroke resulting in disability and death. As stroke outcomes and their significant impacts are unpredictable, improved predictors are needed. This study evaluates the effectiveness and efficiency of machine learning (ML) and deep learning (DL) techniques in predicting stroke risk in different contexts. A systematic review of existing studies and literature was conducted using the Advanced Publications for Systematic Reviews and Meta-analyses (PRISMA) guidelines, focusing on various ML and DL algorithms used for stroke risk prediction. A total of 31 articles met the final inclusion criteria. This review highlights significant advances in stroke prediction with ML and DL models that can handle complex datasets while achieving high prediction accuracy. However, issues related to external validation, standard definition, and transparency remain unresolved. It is recommended to emphasize the importance of features as they can provide insight into the different risk of stroke across countries. The study also shows that the random forest model is the best model for predicting stroke risk, secondary data produces the largest data, and India, including China, and Bangladesh are the countries with the most research on stroke risk. Machine learning and deep learning provide effective ways to predict stroke risk, improving personalized treatment strategies. Solving existing problems is important for their successful integration into treatment.
Cardiovascular Risk Assessment, Neural Network Forecasting, Deep Learning, PRISMA Review, AI in Medical Diagnostics.
Sudhir Kumar Sharma, (2025-01-07 17:48:37.998). Predicting Stroke Risk: Machine Learning Approaches and Their Effectiveness. Abhi International Journal of Applied Science, Volume Yq0ZmvdjjVXBHpcT88Q5, Issue 1.