Cuff-Less Blood Pressure Prediction from ECG and PPG Signals Using Fourier Transformation and Amplitude Randomization...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Continuous blood pressure monitoring is crucial for cardiovascular health management.
• Existing methods often rely on mathematical models or invasive techniques.
• Accurate, non-invasive continuous blood pressure estimation remains a significant challenge.

Purpose of the Study:

• To develop and validate a deep learning algorithm for accurate continuous systolic and diastolic blood pressure monitoring.
• To improve the accuracy of blood pressure estimation using photoplethysmography (PPG) and electrocardiogram (ECG) signals.
• To propose a novel signal preprocessing technique for enhanced deep learning model performance.

Main Methods:

• Acquisition and windowing of PPG and ECG signals at 125 Hz.
• Amplitude randomization and Fourier transformation to extract amplitude and phase information.
• Z-score normalization and supervised learning with a context aggregation network using four input channels.

Main Results:

• Achieved root mean square error (RMSE) of 7 mmHg for systolic blood pressure.
• Achieved RMSE of 6 mmHg for diastolic blood pressure.
• Demonstrated continuous signal prediction without reliance on mathematical models, maintaining measurement performance.

Conclusions:

• The proposed deep learning algorithm offers a highly accurate and non-invasive method for continuous blood pressure monitoring.
• The novel preprocessing technique enhances the ability of deep learning models to interpret PPG and ECG signals.
• This approach represents a significant advancement over traditional mathematical model-based methods for blood pressure estimation.