An Investigation into the Causal Relationship Between Photoplethysmogram and Blood Pressure

Area of Science:

• Physiological signal analysis
• Biomedical engineering
• Causal inference

Background:

• Photoplethysmogram (PPG) signals correlate with blood pressure (BP), but the underlying causal mechanisms are unclear.
• Existing BP estimation methods from PPG are hindered by individual variability, environmental factors, and signal noise.
• Causal inference offers a novel approach to dissecting physiological signal relationships.

Purpose of the Study:

• To investigate the causal relationship between Photoplethysmogram (PPG) components and blood pressure (BP).
• To identify specific PPG frequency components that exhibit a direct causal link with BP.
• To enhance blood pressure estimation accuracy by leveraging causally identified PPG components.

Main Methods:

• Photoplethysmogram (PPG) signal decomposition using ensemble empirical mode decomposition into intrinsic mode functions (IMFs).
• Construction of counterfactual sequences through causal decomposition.
• Causal relationship analysis between IMFs and blood pressure (BP) using the DirectLiNGAM algorithm.

Main Results:

• Intrinsic Mode Functions 4 (IMF4) and 5 (IMF5) demonstrated a significant causal relationship with blood pressure (BP).
• Blood pressure estimation using causally identified PPG components (IMF4, IMF5) yielded lower errors (10.65 mmHg for SBP, 5.75 mmHg for DBP) compared to using the original PPG signal.
• Validation performed on 164 subjects from the VitalDB dataset.

Conclusions:

• Specific intrinsic mode functions (IMFs) of the PPG signal, particularly IMF4 and IMF5, are causally linked to blood pressure (BP).
• Leveraging these causally related PPG components offers a more robust approach for accurate blood pressure estimation.
• This study highlights the potential of causal inference in understanding and improving physiological monitoring techniques.