A Lightweight Deep Learning Model for Fast Electrocardiographic Beats Classification With a Wearable Cardiac Monitor: Development and Validation Study

Abstract

Background Electrocardiographic (ECG) monitors have been widely used for diagnosing cardiac arrhythmias for decades. However, accurate analysis of ECG signals is difficult and time-consuming work because large amounts of beats need to be inspected. In order to enhance ECG beat classification, machine learning and deep learning methods have been studied. However, existing studies have limitations in model rigidity, model complexity, and inference speed. Objective To classify ECG beats effectively and efficiently, we propose a baseline model with recurrent neural networks (RNNs). Furthermore, we also propose a lightweight model with fused RNN for speeding up the prediction time on central processing units (CPUs). Methods We used 48 ECGs from the MIT-BIH (Massachusetts Institute of Technology-Beth Israel Hospital) Arrhythmia Database, and 76 ECGs were collected with S-Patch devices developed by Samsung SDS. We developed both baseline and lightweight models on the MXNet framework. We trained both models on graphics processing units and measured both models’ inference times on CPUs. Results Our models achieved overall beat classification accuracies of 99.72% for the baseline model with RNN and 99.80% for the lightweight model with fused RNN. Moreover, our lightweight model reduced the inference time on CPUs without any loss of accuracy. The inference time for the lightweight model for 24-hour ECGs was 3 minutes, which is 5 times faster than the baseline model. Conclusions Both our baseline and lightweight models achieved cardiologist-level accuracies. Furthermore, our lightweight model is competitive on CPU-based wearable hardware.

Computational Model for Therapy Optimization of Wearable Cardioverter Defibrillator: Shockable Rhythm Detection and Optimal Electrotherapy

“…As per guidelines established by the American heart association (AHA) (Kerber et al, 1997), a sensitivity (Se) higher than 90% for shockable rhythms, and specificity (Sp) higher than 95% for non-shockable rhythms is the benchmark for WCD detection algorithms and our proposed method exceeds the benchmark requirement. Also while comparing with the state-of-the-art, as tabulated in Table 6, our sensitivity-specificity values are closely comparable to the highest accuracy reported by Jeon et al (2020) for WCD applications. Although our reported accuracy is fractionally lower, it is important to note that apart from 8 s standard window-based classification, we have also implemented an overlapping window-based detection that actually spans over 32 s long data that may contain up to 13 windows depending upon the amount of overlapping.…”

“…Some notable prior arts (Silva et al, 2019) implementing CNN architecture for arrhythmia classification, reports the use of various architectural layers (Kwon et al, 2018), attention on noise removal, use of LSTM networks (Krasteva et al, 2020), etc. The most recent work reporting the highest accuracy to date uses a bidirectional LSTM (bi-LSTM) instead of unidirectional LSTM (Jeon et al, 2020).…”

Artificial Intelligence for Detection of Cardiovascular-Related Diseases from Wearable Devices: A Systematic Review and Meta-Analysis

“…After a full text review, 102 studies in total were included in the qualitative review. 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 …”

Electrocardiogram Quality Assessment with a Generalized Deep Learning Model Assisted by Conditional Generative Adversarial Networks

“…In addition to data generation, downsampling is a possible way to assist to train a deep learning model [ 34 ]. Thus, a comparison with a previous quality assessment method [ 35 ] was conducted.…”