Comparative Evaluation of ICG Signal Processing Techniques for aDetecting Caffeine-Induced Hemodynamic Changes

Introduction: The study aims to evaluate the effectiveness of three methods for detecting fiducial points in the impedance cardiography (ICG) signal in the context of subtle hemodynamic changes induced by caffeine intake. The compared approaches include: Time-domain detection (CPDA), extrema-based analysis (EBA) and wavelet transform analysis (WTA). Material and methods: The study involved 20 healthy volunteers, with cardiovascular parameters such as stroke volume (SV), cardiac output (CO) and heart rate (HR) assessed before and after coffee consumption. Results: Comparative analysis showed that WTA provided the lowest variability of hemodynamic parameters, with standard deviations of ±1.54 for CO and ±21.82 for SV, compared with ±1.76 and ±33.77 for EBA and ±1.92 and ±43.59 for CPDA, respectively. Repeated-measures ANOVA revealed significant differences between methods in HR estimation after decaffeinated coffee consumption (p = 0.001). Post-hoc analysis showed significant differences between EBA and CPDA (p = 0.0006), EBA and WTA (p = 0.0003), and EBA and pressure-gauge measurements (p = 0.002). Additional analyses confirmed significant effects of beverage type and method on HR (p = 0.003 for decaffeinated coffee; p = 0.02 for caffeinated coffee) and on CO after decaffeinated coffee (p = 0.03). Conclusions: Overall, the wavelet-based approach demonstrated the greatest robustness to noise and the highest sensitivity to subtle physiological changes associated with caffeine intake. This work highlights the importance of selecting appropriate algorithms for remote cardiovascular monitoring and personalized healthcare and demonstrates their sensitivity to capture subtle changes induced by caffeine ingestion.