PPG2ECGps: An End-to-End Subject-Specific Deep Neural Network Model for Electrocardiogram Reconstruction from Photoplethysmography Signals without Pulse Arrival Time Adjustments

Abstract

Electrocardiograms (ECGs) provide crucial information for evaluating a patient’s cardiovascular health; however, they are not always easily accessible. Photoplethysmography (PPG), a technology commonly used in wearable devices such as smartwatches, has shown promise for constructing ECGs. Several methods have been proposed for ECG reconstruction using PPG signals, but some require signal alignment during the training phase, which is not feasible in real-life settings where ECG signals are not collected at the same time as PPG signals. To address this challenge, we introduce PPG2ECGps, an end-to-end, patient-specific deep-learning neural network utilizing the W-Net architecture. This novel approach enables direct ECG signal reconstruction from PPG signals, eliminating the need for signal alignment. Our experiments show that the proposed model achieves mean values of 0.977 mV for Pearson’s correlation coefficient, 0.037 mV for the root mean square error, and 0.010 mV for the normalized dynamic time-warped distance when comparing reconstructed ECGs to reference ECGs from a dataset of 500 records. As PPG signals are more accessible than ECG signals, our proposed model has significant potential to improve patient monitoring and diagnosis in healthcare settings via wearable devices.

Reconstruction of Missing Electrocardiography Signals from Photoplethysmography Data Using Deep Neural Network

“…The accuracy of these methods depends on the accuracy of the R wave in ECG and contraction seam extraction algorithms in PPG, which can reduce the accuracy of ECG reconstruction. The computational parametric model [8], lightweight neural network [9], deep learning models based on encoder-decoder [10], BiLSTM [11], PPG2ECGps [12], P2E-WGAN [13], CardioGAN [14], Performer [15], transformed attentional neural network [16], and banded kernel ensemble method [17] have been proposed for reconstructing electrocardiograms from PPG based on deep learning methods. In [8], the author proposed a computational parametric model that extracts features from PPG to predict ECG parameters.…”

“…However, cycle alignment and segmentation result in loss of temporal information, such as pulse transit time and heart rate variability, which are essential clinical factors. Some studies [11][12][13][14][15][16] used segment reconstruction of ECG signals from PPG as a basis. The models proposed in the first two studies targeted specific subjects and could not be generalized to multiple subjects, representing a limitation.…”

UNet-BiLSTM: A Deep Learning Method for Reconstructing Electrocardiography from Photoplethysmography

“…Several studies have used PPG signals to reconstruct ECG signals using various techniques. Several studies have utilized the discrete cosine transform (DCT) method [9], crossdomain joint dictionary learning (XDJDL) method [10], lightweight neural networks [11], bidirectional long short-term memory (BiLSTM) models [12], and the PPG2ECGps model [13] to reconstruct ECG signals for subject-specific models. The first two studies proposed reconstructing ECG from PPG using a mathematical model.…”

Biomedical Data Mining and Machine Learning for Disease Diagnosis and Health Informatics

“…By integrating innovative biomedical data mining and machine learning technologies, smart healthcare systems can provide more accurate and efficient diagnostic solutions [7] and treatment opportunities [8], aiming towards the purpose of significantly improving overall healthcare quality and rehabilitation outcomes [9]. In this Special Issue, we strive to highlight the recent development of biomedical data mining and machine learning technologies for the diagnosis of infectious diseases and chronic diseases; the topics also cover the theoretical advances and practical ap-plications of deep learning neural network architectures for physiological signal measurement and data analysis [10].…”