Have a personal or library account? Click to login
Filter and Sampling Rate Optimization for PPG-Based Detection of Autonomic Dysfunction: An ECG-guided Approach Cover

Filter and Sampling Rate Optimization for PPG-Based Detection of Autonomic Dysfunction: An ECG-guided Approach

Open Access
|Aug 2025

Figures & Tables

Fig. 1.

The placement of the ECG electrodes with PowerLab, a physiological signal acquisition instruments from ADInstruments Inc. [30].
The placement of the ECG electrodes with PowerLab, a physiological signal acquisition instruments from ADInstruments Inc. [30].

Fig. 2.

Examples of ECG and PPG signals.
Examples of ECG and PPG signals.

Fig. 3.

Signal processing flow chart for ECG and PPG.
Signal processing flow chart for ECG and PPG.

Fig. 4.

The process of extracting QRS from raw signals.
The process of extracting QRS from raw signals.

Fig. 5.

Detected ECG and PPG peaks.
Detected ECG and PPG peaks.

Fig. 6.

HeatMap of correlation coefficients between the RRI at 1 kHz and the SSIs at various rates in resting (a), breathing phase (b).
HeatMap of correlation coefficients between the RRI at 1 kHz and the SSIs at various rates in resting (a), breathing phase (b).

Fig. 7.

Evaluation of the best SSIV assessment range of normal cardiac autonomic function with the ROC curve, (a) – the resting phase; (b) – the deep breathing phase.
Evaluation of the best SSIV assessment range of normal cardiac autonomic function with the ROC curve, (a) – the resting phase; (b) – the deep breathing phase.

Fig. 8.

HeatMap of average absolute errors between the RRI at 1000 Hz and the SSIs at different sampling rates in resting (a) and breathing phases (b).
HeatMap of average absolute errors between the RRI at 1000 Hz and the SSIs at different sampling rates in resting (a) and breathing phases (b).

Confusion matrix between RRIV and SSIV in the resting and deep breathing phases_

PhaseSSIVRRIV Abnormal [ms]RRIV Normal [ms]
RestingAbnormal442
Normal8190
Deep breathingAbnormal443
Normal7190

SSIV assessment range of normal cardiac autonomic function in each age group based on PPG_

Age [y]20 ∼ 2930 ∼ 3940 ∼ 4950 ∼ 5960 ∼ 69
Rest [ms]13 ∼ 457 ∼ 317 ∼ 356 ∼ 228 ∼ 18
Breath [ms]20 ∼ 6110 ∼ 5315 ∼ 4712 ∼ 589 ∼ 27

The sensitivity and specificity of SSIV in different assessment ranges_

PhaseRate [%]Out+2Out+1RRIV sourceIn-1In-2In-3In-4
RestSensitivity36.5463.4684.6296.1596.15100.00100.00
Specificity98.9698.9698.9691.6779.6970.8359.90
BreathSensitivity54.9070.5986.2794.1294.1296.0896.08
Specificity99.4899.4898.4595.8591.1984.9778.24

A descriptive listing of PPG signal processing methods_

No.Filter nameDescription
F01ButterworthThe passband frequency is from 1 Hz to 5 Hz, and the filter order is 2.
F02Bessel
F03Chebyshev
F04Elliptic

F11Savitzky-GolayThe polynomial order is 3, and the window size 401.
F12AverageThe passband frequency is from 1 Hz to 5 Hz.
F13Periodic movingAfter normalizing each cardiac cycle with the same number of sampled signals, the ith sampled signals of each of the 5 cycles are averaged, where 1 ≤ i ≤ the normalized cycle length.

F21DaubechiesUse the inverse DWT of Daubechies db6 and Coiflet C3 to reconstruct the selected band from 1 Hz to 5 Hz.
F22Coiflet C3
F23Morlet (CWT)Apply the inverse WT of Morlet to reconstruct the selected band from 1 Hz to 5 Hz.

RRIV assessment range of normal cardiac autonomic function in each age group based on ECG [38]_

Age [y]20 ∼ 2930 ∼ 3940 ∼ 4950 ∼ 5960 ∼ 69
Rest [ms]12 ∼ 466 ∼ 326 ∼ 365 ∼ 237 ∼ 19
Breath [ms]19 ∼ 629 ∼ 5414 ∼ 4811 ∼ 598 ∼ 28

Average absolute errors between the RRIV/ECG SDNN/ECG, NN50 at 1000 Hz and the SSIV/PPG SDNN/PPG NN50 at different sampling rates_

Filter/Sample ratesPhaseParameter1000 [Hz]500 [Hz]250 [Hz]200 [Hz]125 [Hz]100 [Hz]50 [Hz]40 [Hz]25 [Hz]20 [Hz]
F01RSSIV [%]0.60.61.21.31.31.41.82.23.44.4
F11RSSIV [%]1.41.52.12.22.32.32.834.17.3
F23RSSIV [%]22.232.52.92.83.24.74.65.8
F01BSSIV [%]0.91.11.21.11.21.21.822.93.8
F11BSSIV [%]2.52.32.32.52.62.73.13.34.214.4
F23BSSIV [%]2.62.72.92.52.933.454.35.6

F01RSDNN [ms]0.60.60.60.60.80.92.13.16.59.5
F11RSDNN [ms]1.31.41.51.61.71.93.247.312.4
F23RSDNN [ms]2.72.62.82.12.72.33.46.97.510.8
F01BSDNN [ms]1.21.21.31.31.31.42.22.757.2
F11BSDNN [ms]2.22.22.22.42.42.63.33.96.315.3
F23BSDNN [ms]2.72.72.92.42.72.63.36.15.68.1

F01RNN50 [%]0.80.90.8111.31.71.73.63.1
F11RNN50 [%]1.71.61.81.61.923.52.25.14.3
F23RNN50 [%]1.51.51.51.61.61.82.22.14.13.3
F01BNN50 [%]0.80.80.80.90.91.12.21.83.63.1
F11BNN50 [%]2.42.42.82.32.72.34.92.85.95.3
F23BNN50 [%]1.61.61.71.61.71.93.12.54.84.1
Language: English
Page range: 200 - 211
Submitted on: Oct 8, 2024
|
Accepted on: Jul 2, 2025
|
Published on: Aug 28, 2025
In partnership with: Paradigm Publishing Services
Publication frequency: Volume open

© 2025 Yi-Hui Kao, Danyal Shahmirzadi, Sung-Tsang Hsieh, Wen-Fong Wang, published by Slovak Academy of Sciences, Institute of Measurement Science
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.