Causal inference based cuffless blood pressure estimation: A pilot study.

Area of Science:

• Biomedical Engineering
• Data Science
• Cardiovascular Health

Background:

• Wearable sensors like photoplethysmography (PPG) and electrocardiography (ECG) enable cuffless blood pressure (BP) monitoring.
• Current data-driven methods often overlook causal relationships, focusing on correlations.

Purpose of the Study:

• To investigate the feasibility of applying causal inference for cuffless BP estimation.
• To identify wearable sensor features causally linked to BP variations.

Main Methods:

• Utilized the Fast Causal Inference (FCI) algorithm to identify causal graphs and features.
• Integrated causal inference mechanisms into BP estimation models using time-lagged links.
• Validated the approach on 62 subjects with continuous ECG, PPG, and BP data.

Main Results:

• Discovered novel causal features superior to Pulse Transit Time (PTT) for tracking BP changes.
• Achieved a Mean Absolute Difference (MAD) of 5.10 mmHg for systolic BP (SBP) and 2.85 mmHg for diastolic BP (DBP).
• Demonstrated superior performance compared to traditional models that do not account for causality.

Conclusions:

• This research pioneers the application of causal inference in cuffless BP estimation.
• The findings offer insights into the mechanisms, methodologies, and applications of cuffless BP monitoring.