Utilizing machine learning approaches for the estimation and detection of atrial fibrillation, a prevalent rhythm disease having substantial clinical consequences, especially associated with a higher risk of ischemic cerebral ischemia and cardiac arrest, has sparked a lot of attention recently. Earlier studies have identified many clinical indicators that may be used to anticipate the onset of atrial fibrillation prior to the introduction of artificial intelligence in healthcare delivery. Prior diagnostics, laboratory findings, imaging data, and electrophysiological measurements are all examples of clinical features. The electronic medical record has a lot of information like this, and artificial intelligence systems may query it automatically. We discuss the latest state of machine learning methods for atrial fibrillation detection and prediction, as well as the consequences and future outlook for this rapidly developing area, using the technologically advanced computational capabilities.