Described herein are user-wearable devices, and methods for use therewith, for monitoring for one or more types of arrhythmias based on a photoplethysmography (PPG) signal obtained using an optical sensor of a user-wearable device. A PPG based statistical and/or machine learning model is used to analyze a PPG signal, obtained using the optical sensor, to monitor for one or more types of arrhythmias including atrial fibrillation (AF). In response to detecting an arrhythmia based on the PPG signal, an electrocardiogram (ECG) signal is obtained using an ECG sensor of the user-wearable device. An ECG based statistical and/or machine learning model is used to analyze the ECG signal obtained using the ECG sensor of the user-wearable device to confirm or reject the arrhythmia detected based on the PPG signal and/or to perform arrhythmia discrimination. Obtained PPG and/or ECG signal segments can be provided to the model(s) to update the model(s).
