Key result
Novel ECG identification method using light CNN achieves ~99% authentication accuracy for wearable devices.
Why the study?
ECG signals in wearable devices are easily affected by emotional heart rate, while traditional identification schemes are complicated with low authentication accuracy.
A novel lightweight CNN-based ECG identification method using GASF graphs achieves high authentication accuracy (99.26%) suitable for resource-constrained wearable devices.
Promising accuracy supports ECG biometrics on wearables; leaves open real-world validation before clinical use.
In order to solve the problem that the electrocardiogram(ECG) signal under the wearable device is easily affected by the emotional heart rate, the traditional identification scheme is complicated and the authentication accuracy is low. In this paper, a novel ECG identification method combining Gramian Angular Summation Field (GASF) graph and two-dimensional light convolutional neural network is proposed. This method can solve the shortcomings of manual feature extr-action in traditional schemes and enhance the accuracy of authentication recognition by direct application of large complex one-dimensional convolutional neural network, and can be better applied to wearable devices. In the experiment, first of all, considering that the ECG signal is easily affected by the heart rate and emotion when the user uses the wearable device, this paper selects a recording segment centered on the QRS complex with a size of 250 sample points as the input data set in the selection of the data set, to enhance the accuracy and robustness of authentication. Then the one-dimensional time series is converted into a two-dimensional image using the GASF method as the input of the neural network. On this basis, according to the resource-constrained characteristics of wearable devices, a set of light neural network architecture with fewer parameters, higher efficiency and more suitable for wearable devices is designed using the idea of depth separable convolution. . Finally, compared with the performance indicators of previous research schemes, the method proposed in this paper has a more significant effect, and its best accuracy value is 99.26%.
No takes yet. Share an insight, caveat, or question.
He et al. (2023) studied ECG identification. Gramian Angular Summation Field (GASF) graph and two-dimensional light convolutional neural network vs. Traditional identification schemes was evaluated on Authentication accuracy. A novel ECG identification method combining Gramian Angular Summation Field graphs and a light convolutional neural network achieved a best authentication accuracy of 99.26% for wearable devices.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: