Comparative assessment of AD8232- and MAX30003-based ECG acquisition paths using synthetic ECG signals
Abstract
Purpose
This study presents a comparative technical evaluation of two implemented electrocardiogram (ECG) acquisition paths based on AD8232 and MAX30003 under controlled laboratory conditions. The analog front-end paths were designed in accordance with the reference circuits and design recommendations specified in the manufacturer’s documentation for the AD8232 and MAX30003 devices.
Methods
Synthetic ECG signals were generated using a patient simulator for heart rates of 30, 60, 80, and 100 beats per minute and were recorded simultaneously from both acquisition paths at a sampling frequency of 128 Hz. The processing pipeline comprised time-axis standardization, baseline correction, local cross-correlation-based synchronization, linear amplitude alignment, QRS detection using a simplified Pan–Tompkins-inspired procedure, and median-centered amplitude standardization. Signal similarity after preprocessing was assessed using time-domain metrics, including Pearson correlation, mean absolute error, root mean square error (RMSE), and normalized RMSE. In addition, spectral analysis based on Welch’s method was performed using absolute bandpower values and relative bandpower indices.
Results
After synchronization and normalization, both acquisition paths showed very high morphological agreement, with Pearson correlation exceeding 0.99 across all tested heart-rate settings. Error metrics remained low and comparable between recordings. Spectral analysis revealed marked differences in absolute power, reflecting properties of the complete acquisition chains rather than signal quality alone. Relative spectral indices showed implementation-specific differences between the complete acquisition paths, particularly in the 50 Hz component band, whereas the 40–60 Hz band was retained only as an auxiliary descriptive upper-band index because of the different filtering and digitization architectures of the two paths.
Conclusions
Under the tested synthetic and controlled conditions, both implemented ECG acquisition paths provided comparable post-processed morphological representations of the ECG signal, while differing in spectral characteristics. The examined paths differed in the relative distribution of spectral energy under the applied configuration. However, these findings should be interpreted as implementation-specific technical observations rather than evidence of general device superiority. Further validation in human recordings and real-world conditions is required.
© 2026 Fryderyk Gajdzik, Michał Kimbar, Paweł Krowicki, published by Wroclaw University of Science and Technology
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.