Cardiovascular Stroke is a potentially fatal condition that requires careful attention to achieve the best results. Early detection and intervention are crucial to reduce mortality and long-term damage associated with stroke. This study provides new solutions to these urgent needs by offering instant stroke detection based on deep learning and machine learning to increase performance. The most important goal of this research is to create a reliable and accurate system that can instantly distinguish strokes, allowing doctors to make informed decisions. Traditional stroke diagnosis technology relies on the interpretation of medical images, which is laborious and prone to human error. Although ML and DL approaches have demonstrated potential for automating this procedure, difficulties still exist since large and varied datasets are required. In order to overcome these obstacles, we suggest YOLOv9 + Capsnet in conjunction with XG-BOOST algorithms on large-scale datasets that include stroke and non-stroke cases according to the cardiovascular disease of the subjects seen in the pictures. Through this process of training, the model gains the ability to recognise complex patterns and features related to strokes, which improves the accuracy of its diagnosis. With the ability to increase patient caring and diagnosis of strokes, this optimised model has great promise for saving lives and raising the standard of healthcare. This article aims to identify the most effective and practical machine learning and deep learning algorithms for the prediction of cardiac illnesses by summarising some of the recent research in this area. Prediction systems’ future directions have also been taken into account.