Assessment of Potential Primary and Recurrent Ischemic Stroke by Detecting Atrial Fibrillation using 1D-CNN and CHA2DS2-VA Score

Among all kinds of arrhythmia, atrial fibrillation is the most frequently handled in the medical system with steep gradients with increasing age and is responsible for ischemic stroke. The main challenge is an adverse consequence of AF which might happen even when the AF has even detected inpatient, on the other hand, AF may occur for the first time even after the first stroke which might be responsible for recurrent stroke during Neurorehabilitation. Due to the advantage of the 1D-CNN algorithms such as inexpensive hardware and the ability to implement in real-time using compact architecture, this is becoming popular with personalized biomedical data classification. This paper presents an approach using the 1D- CNN algorithm to automatically detect AF in ECG signals from a patient to further use that knowledge and assess stroke management system involving a scoring system. Primarily, the performance of the proposed method was trained and validated using an arrhythmia dataset from physionet and showed an average accuracy of 96% with a loss of less than.15.