Aims: Photoplethysmography (PPG) is a non-invasive biosignal that optically measures changes in microvascular blood volume, reflecting car-diovascular information including heart rate (HR), stroke volume (SV), and vascular properties. PPG morphological features are quantitative indi-ces of waveform shape that reflect hemodynamic status. However, it re-mains unclear whether these features predominantly reflect cardiac factors such as HR and SV, or whether they are also confounded by arterial prop-erties. This study aimed to identify features that more predominantly re-flect HR or SV changes accounting for measurement site effects.
Methods: Using an in silico hemodynamic model based on the Pulse Wave Database (PWDB), HR and SV were each varied across 200 incre-mental steps while other cardiovascular parameters were held constant. PPG signals were simulated at 25 arterial segments, including the brachial, radial, and digital arteries, and 40 morphological features were extracted. Correlation analyses were performed on digital PPG, and features with both |Pearson r| and |Spearman ρ| ≥ 0.8 were selected as candidate features. Parameter-by-segment interaction analysis was then conducted to evaluate measurement site effects. Inter-segment variability was quantified using the coefficient of variation (CV). This yielded 39 of 40 candidate features for HR and 22 of 40 for SV. Significant parameter-by-segment interactions were observed for all candidate features, indicating susceptibility to meas-urement site effects.
Results: CV analysis revealed that time-based features—t_dia (CV = 0.0055) and t_sys (CV = 0.0273)—showed low inter-segment variability for HR. For SV, features related to systolic duration and area—t_sys (CV = 0.0227), t_dia (CV = 0.0227), and A1 (CV = 0.1528)—showed similarly low inter-segment variability. Although all candidate features were influ-enced by measurement site, the magnitude of this influence varied across features.
Conclusion: Time-based and systolic features showed lower inter-segment variability, suggesting more consistent reflection of HR or SV changes regardless of measurement site.