Speech Title: Big Data Processing and Machine Learning for Clinical Atrial Fibrillation Detection
Abstract: Atrial fibrillation (AF) is a tachyarrhythmia characterized by predominantly uncoordinated atrial activation with consequent deterioration of atrial mechanical function. AF is the most common sustained cardiac arrhythmia, occurring in 1-2% of the general population, and is associated with significant mortality and morbidity. Despite the widespread of AF disease in clinical practice, the detection for AF still remains challenging, especially when processing the dynamic ‘Big ECG’ data from the real-time acquirement. This talk will present the discussion about two types of automatic AF detection: atrial activity analysis-based or ventricular response analysis-based methods. For dynamic ECG processing, ventricular response analysis-based method seems like more suitable since the robustness to a variety of noise sources. Thus, the talk will focus on the discussion of AF features from the ventricular response analysis. PhysioNet/Computing in Cardiology Challenge 2017 posted a large number of short-term AF/non-AF ECG recordings, and different big data processing and machine learning methods will be analyzed/tested based on this database. We aimed to identify the key steps for accurate automatic AF detection for clinical application.
Keywords: atrial fibrillation, arrhythmia, ECG, big data, machine learning