Research on Heart Rate Estimation Algorithm for PPG Signals Based on SVMD-VP

Photoplethysmographic signals (PPG) can accurately estimate an individual's heart rate (HR), and its accuracy is a crucial measurement indicator for wearable devices. To address the issue of HR estimation accuracy in PPG, a method based on successive variational mode decomposition integrated with variational mode decomposition-principal component analysis (SVMD-VP) is proposed. The intrinsic mode components (IMFs) from the PPG signal are extracted using SVMD. The extracted IMFs are then selectively reconstructed through Fast Fourier Transform (FFT). The VMD-PCA algorithm is applied to further decompose and reconstruct the signal, and the IMF corresponding to the first principal component (PC1), which is most influenced by HR, is selected to extract the HR. The algorithm validates the accuracy of HR estimation using PPG data of different data lengths (EL) from a self-collected database. The experimental results indicate that the algorithm achieves the highest accuracy in HR estimation at <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{E L}=\mathbf{3 0 s}$</tex>. For the healthy elderly group, the mean absolute error (MAE), mean percentage error (PE), and root mean square error (RMSE) are 0.338 bpm, 0.005 %, and 0.535 bpm, respectively, resulting in a Pearson correlation coefficient (PCC) of 0.998. For the diabetic patient group (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{EL}=30 ~\mathrm{s}$</tex>), the MAE, PE, and RMSE are <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$0.278 \text{bpm}, 0.004 \%$</tex>, and 0.475 bpm, respectively, achieving a PCC of 0.999. The SVMD integrated VMD-PCA algorithm can estimate the HR from PPG with different ELs in two distinct populations, and the estimated HR values are optimal compared to those obtained using the EMD and VMD fusion algorithms.