Objective: Photoplethysmogram (PPG) signals have been widely used to estimate blood pressure (BP) cufflessly and continuously. A number of different PPG features have been proposed and extracted from PPG signals with the aim of accurately estimating BP. However, the underlying physiological mechanisms of PPG-based BP estimation still remain unclear, particularly those corresponding to various PPG features. In this study, the physiological mechanisms of PPG features for BP estimation were investigated, which may provide further insight.
Approach: Experiments with cold stimuli and an exercise trial were designed to change the total peripheral vascular resistance (TPR) and cardiac output (CO), respectively. Instantaneous BP and continuous PPG signals from 12 healthy subjects were recorded throughout the experiments. A total of 65 PPG features were extracted from the original, the first derivative, and the second derivative waves of PPG. The significance of the change of PPG features in the cold stimuli phase and in the early exercise recovery period was compared with that in the baseline phase.
Main results: Intensity-specific PPG features changed significantly (p < 0.05) in the cold stimuli phase compared with the baseline phase, demonstrating that they were TPR-correlated. Time-specific PPG features changed significantly (p < 0.05) in the early exercise recovery period compared with the baseline phase, suggesting they were CO-correlated. Most of the PPG features associated with slope and area changed obviously both in the cold stimuli phase and in the early exercise recovery period, indicating that they should be TPR-correlated and CO-correlated.
Significance: The findings of this study explained the intrinsic physiological mechanisms underlying PPG features used for BP estimation, and provided insights for exploring more diagnostic applications of the PPG features.