AFibNet: an implementation of atrial fibrillation detection with convolutional neural network

Abstract

Background Generalization model capacity of deep learning (DL) approach for atrial fibrillation (AF) detection remains lacking. It can be seen from previous researches, the DL model formation used only a single frequency sampling of the specific device. Besides, each electrocardiogram (ECG) acquisition dataset produces a different length and sampling frequency to ensure sufficient precision of the R–R intervals to determine the heart rate variability (HRV). An accurate HRV is the gold standard for predicting the AF condition; therefore, a current challenge is to determine whether a DL approach can be used to analyze raw ECG data in a broad range of devices. This paper demonstrates powerful results for end-to-end implementation of AF detection based on a convolutional neural network (AFibNet). The method used a single learning system without considering the variety of signal lengths and frequency samplings. For implementation, the AFibNet is processed with a computational cloud-based DL approach. This study utilized a one-dimension convolutional neural networks (1D-CNNs) model for 11,842 subjects. It was trained and validated with 8232 records based on three datasets and tested with 3610 records based on eight datasets. The predicted results, when compared with the diagnosis results indicated by human practitioners, showed a 99.80% accuracy, sensitivity, and specificity. Result Meanwhile, when tested using unseen data, the AF detection reaches 98.94% accuracy, 98.97% sensitivity, and 98.97% specificity at a sample period of 0.02 seconds using the DL Cloud System. To improve the confidence of the AFibNet model, it also validated with 18 arrhythmias condition defined as Non-AF-class. Thus, the data is increased from 11,842 to 26,349 instances for three-class, i.e., Normal sinus (N), AF and Non-AF. The result found 96.36% accuracy, 93.65% sensitivity, and 96.92% specificity. Conclusion These findings demonstrate that the proposed approach can use unknown data to derive feature maps and reliably detect the AF periods. We have found that our cloud-DL system is suitable for practical deployment

Energy efficient convolutional neural networks for arrhythmia detection

“…The filtering step could be replaced by some extra layers of convolutions, but as the main idea of our implementation is to maintain a small network with as few neurons and parameters as possible, the band-pass filter was essential for noise canceling. While some previous works apply in a similar manner filtering of small and high frequencies [26], the transformation to the frequency domain by a Fourier or wavelet transform is also used [10]. With the intention of incorporating our designed model on a physical chip, the band-pass filter can offer an "inexpensive" solution, given that it can be applied directly in time domain, avoiding this way extra transformations and it is a well established method for analog and digital chips.…”

“…Additionally, we tried to generate the 1D-CNN as it is described in [25] and [26] for atrial fibrillation and trained it on our data. However, it was not feasible to achieve model convergence and produce a stable accurate solution.…”

Deep learning-based electrocardiogram rhythm and beat features for heart abnormality classification

“…Using this approach, pattern abnormalities that occurred in the ECG, both in beats and rhythms, can be detected only using single architecture. We generalized the 1D-CNN architecture that was published in previous work ( Nurmaini et al, 2020 ; Tutuko et al, 2021 ). The proposed methodology of the 1D-CNN generalized architecture is presented in Fig.…”

“…DL-based frameworks mainly include a stacked autoencoder (SAE), long short-term memory (LSTM), a deep belief network (DBN), convolutional neural networks (CNN), and so on. Among DL algorithms, we have generated a one-dimensional CNN (1D-CNN) model and showed promising results in our previous works ( Nurmaini et al, 2020 ; Tutuko et al, 2021 ). In other works, 1D-CNN has also performed well for ECG classification, with overall performances ranging from 93.53% to 97.4% accuracy using rhythm ( Acharya et al, 2017 ; Wang, 2020 ) and with overall 92.7% to 96.4% accuracy using beat ( Zubair, Kim & Yoon, 2016 ; Kiranyaz, Ince & Gabbouj, 2015 ).…”

Heartbeat Classification and Arrhythmia Detection Using a Multi-Model Deep-Learning Technique

“…To accomplish the aforementioned tasks, neural networks such as Recurrent Neural Networks (RNN) [ 20 , 21 , 22 ], Long Short-Term Memory (LSTM) [ 23 , 24 ], Convolutional Neural Networks (CNN) [ 25 , 26 , 27 ], as well as hybrid models [ 28 , 29 , 30 , 31 ] etc., are being integrated to overcome the hindrances of conventional machine-learning strategies that were subject to manual and inaccurate selection of features that may incite inconvenient impacts for the current applications. The drawbacks of the hybrid approaches accumulate the increasing cost and lack of quality datasets which, however, can be considered negligible in some viable cases because the precise classification of heartbeats along with the accurate detection of arrhythmia requires a substantial amount of data to work with [ 32 ].…”