Background: Atrial fibrillation (AFIB) is one of the most common types of arrhythmia, which leads to heart failure and stroke to public. As AFIB has the high potential to cause permanent disability in patients, its early detection is extremely important. There are different types of AFIB classification algorithm that have been proposed by researchers in recent years. Methods: This paper reviews the features of AFIB in terms of ECG morphological features and heart rate variability (HRV) analysis on different methods. The existing classification method, particularly focusing on Artificial Intelligence technique, is also comprehensively described. Other than that, the existing implementation technology of arrhythmia detection platforms such as smart phone and System-on-Chip-based embedded device are also elaborated in terms of their design trade-offs. Conclusion: Current existing AFIB detection algorithm cannot compromise for high accuracy and low complexity. Due to the limitation of embedded system, design trade off should be considered to strike the balance between the performance of algorithm and the limitation.

Stroke is one of the most severe cardiovascular disease which can potentially cause permanent disability. Atrial Fibrillation (AF) is one of the major risk factors of stroke that can be detected from electrocardiogram (ECG) monitoring.  Objective This study proposed an AF detection algorithm based on stationary wavelet transform (SWT) and artificial neural network (ANN) for screening purpose. The algorithm is aimed for embedded System-on-Chip (SoC) technology deployment as a standalone AF classifier for community in rural area where the internet infrastructure may not well established. Methods After standard ECG signal pre-processing, SWT is applied to filtered ECG and produces 12 sets of primary features in time-frequency domain. The power spectral density (PSD) and log energy entropy (LogEn) were calculated from these 12 sets of primary features, to measure atrial activity fall in frequency range of 4 to 9 Hz, and the randomness of an ECG signal caused by AF, respectively. Finally, the ANN classifier recognizes the pattern of AF based on high atrial activity and randomness of ECG signal. Algorithm exploration is carried out to determine the optimum parameter value which can yield the best classification and suitable to be implemented in embedded SoC technology for real-time computation performance. ECG training and testing datasets of the proposed AF detection algorithm were extracted from MIT-BIH Atrial Fibrillation Database which consists of 23 ECG record with each record contains a 10 hours ECG data.  Results AF detection accuracy is 95.3% which was able to classify an ECG signal into categories of AF, sinus rhythm, and other arrhythmia.  Conclusion The proposed AF detection algorithm based on combination of SWT and ANN can achieve high accuracy and is suitable to be implemented as a standalone AF classifier based on embedded SoC technology targeted for early detection of AF in the community.

No abstract