Prolonged Sleep Apnea is a sleeping disorder that can cause arrhythmia, hypertension, and other serious health conditions leading to cardiovascular diseases and fatal strokes. Most widely used current clinical techniques for sleep apnea diagnosis are expensive, time-consuming, and cannot be performed remotely. Wearable watch-style health trackers continuously track sleep behavior, physiological data, and physical activity that can enable real-time remote diagnosis of sleep apnea. Recently, the application of Artificial Intelligence (AI) techniques within the field of medicine and remote diagnosis is gaining popularity. In this paper, several Artificial Intelligence (AI) models have been trained and tested to classify sleep apnea condition in real-time using sequential data of Instantaneous Heart Rates (IHR). Using the confusion matrix, the accuracy, precision, recall, specificity, Fl Score, sensitivity, and area under the receiver operating characteristic curve of each model are computed and compared. The Bi-directional Long Short-Term Memory (LSTM) was found to be the best AI technique for classifying sleep apnea. The approach depicted in this study for diagnosing sleep apnea can allow the telemedicine, telehealth, and mHealth applications to detect several health risk factors in real-time using data streaming from the health trackers.