Atrial fibrillation(termed as AF/Afib henceforth) is a discrete and often rapid heart rhythm that can lead to clots near the heart. We can detect Afib by ECG signal by the absence of p and inconsistent intervals between R waves as shown in fig(1). Existing methods revolve around CNN that are used to detect afib but most of them work with 12 point lead ECG data where in our case the health gauge watch deals with single-point ECG data. Twelve-point lead ECG data is more accurate than a single point. Furthermore, the health gauge watch data is much noisier. Implementing a model to detect Afib for the watch is a test of how the CNN is changed/modified to work with real life data