A stroke can cause a variety of problems for society as a whole in addition to its finances, and it is a significant contributor to infirmity in both individuals and the elderly. These problems can be caused by the fact that a stroke is a major contributor to infirmity. The majority of patients diagnosed with stroke have abnormal biosignals, such as an abnormal electrocardiogram (ECG). It is essential to accurately measure and monitor the biosignals of each individual patient in order to administer treatment in a timely manner. This can only be accomplished by treating each patient as an individual. However, the vast majority of the time and resources spent on developing costly and complicated image-processing technologies are invested in developing attack diagnosis and prediction systems. This study intends to develop an artificially intelligent system that is capable of predicting cerebrovascular accidents by incorporating into the analysis process biosignals obtained from the electrocardiogram (ECG) and the electroencephalogram (PPG). However, previous research has concentrated more on developing criteria for clinical or immediate therapy following the initial symptoms of a stroke rather than identifying the early warning signs of a stroke. Instead of making an attempt to diagnose a stroke in its earliest stages, this was done instead. In this study, we propose a method that is based on machine learning and has the ability to predict as well as semantically interpret strokes by making use of multiple modalities of biosignals. Given that deep learning techniques appear to be more accurate than machine learning algorithms, we have considered a Naive Bayes algorithm with an accuracy rate of 93%,LSTM with an accuracy rate of 90%.