An ECG–PPG Physiological Signal Emulator for Calibration and Validation of Cardiovascular Monitoring Devices

Figures

• 

  Figure 1. Architectural diagram and overall operation flowchart of a synchronized ECG–PPG signal generation system using an OEM STM32F407VET6-based development board (STM32F407VET6 microcontroller, STMicroelectronics, Geneva, Switzerland).

• 

  Figure 2. Block diagram of the electrophysiological signal simulation model.

• 

  Figure 3. Schematic diagram of a signal channel: (a) DAC circuit, (b) LPF circuit, and (c) buffer circuit. Colors follow the default EasyEDA Pro schematic convention: green lines indicate electrical connections, red symbols represent circuit components and power/ground terminals, and blue text denotes signal/net and component labels. The schematic was drawn using EasyEDA Pro version 2.2.45.

• 

  Figure 4. Morphological and vectorcardiographic characterization of the synthesized 12-lead ECG under normal sinus rhythm at 75 bpm and an amplitude scaling of 1 mV: (a) representative ECG waveforms from leads II, aVR, V1, and V5, illustrating rhythm, polarity, and precordial morphology, (b) R-wave height and S-wave depth across the standard 12 leads, and (c) vectorcardiographic loops of the P wave, QRS complex, and T wave projected onto the frontal, transverse, and left sagittal planes.

• 

  Figure 5. Conduction behavior of the block-based rhythm generator: (a) Lead II waveforms for the six implemented rhythms, with triangles marking QRS complexes, (b) progressive PR prolongation within each Mobitz I cycle at 75 bpm and 1 mV, comparing the prescribed values with those reported by the Kenz Cardico 306, and (c) agreement between direct and RR-derived PR measurements across the evaluated heart-rate range at 1 mV.

• 

  Figure 6. Rhythm-dependent electro-mechano-vascular coupling: (a) simultaneous ECG and PPG waveforms acquired on two oscilloscope channels at the analog buffer outputs, for sinus rhythm at 75 bpm and 2 mV (top), and ventricular tachycardia at 140 bpm and 2 mV (bottom); (b) beat-level distribution of the PAT for the two rhythms.

• 

  Figure 7. Characteristics of the analog circuit sequence: (a) amplitude and phase response of the ECG channel; (b) amplitude and phase response of the PPG channel; (c) amplitude error as a function of the set voltage, with the inset showing the error at the DAC, low-pass filter, and buffer at 1000 mV; and (d) THD and SNR as a function of the set voltage for both channels.

• 

  Figure 8. Inter-channel timing and playback stability of the ECG output system: beat-to-beat channel skew, Δtskew, between the LL and RA channels measured from common QRS complexes. Black dots connected by a line are the per-beat values; the red line and shaded band are the mean and its 95% confidence interval.

• 

  Figure 9. Playback stability of the ECG output across set heart rates: (a) RR-interval scatter, (b) waveform autocorrelation at the playback-buffer length, and (c) R-peak amplitude coefficient of variation, each at set rates of 40–120 bpm and three signal sensitivities (0.5, 1.0, 2.0 mV).

• 

  Figure 10. Output stability during 60 min of continuous operation. Deviations in PAT, R-peak amplitude, PPG amplitude, and ECG DC offset from their respective session means were evaluated at 5-min intervals. No significant drift was observed in any metric, and the shaded region indicates the ±1% range.

• 

  Figure 11. End-to-end reproduction of a clinical 12-lead recording: (a) the digital reference and recorded waveforms for four representative leads, namely II, aVR, V1, and V5; (b) the transfer gain across all twelve leads together with the corresponding frequency distribution of reference-signal energy; and (c) the decomposition of the reproduction error into a fixed noise floor and an amplitude-proportional component.

• 

  Figure 12. Verification of the emitted ECG signal using the Kenz Cardico 306 electrocardiograph: (a) recorder-reported heart rate versus the programmed rate from 40 to 120 bpm at sensitivity settings of 0.5, 1.0, and 2.0 mV, with the identity line indicating ideal agreement, while the inset shows the residual relative to the realizable rate imposed by integer-length buffer quantization; (b) representative lead II beats at the three sensitivity settings after normalization to the R-wave peak, showing preservation of waveform morphology, with the inset presenting the corresponding waveforms before normalization on the absolute amplitude scale; and (c) reported RV5 and SV1 amplitudes as functions of the sensitivity setting, where thin gray lines represent individual rate levels, the thick black line represents the mean across rates, and the dashed line indicates the ideal proportional response through the origin.