A Novel Neural Network Model for Blood Pressure Estimation Using Photoplethesmography without Electrocardiogram.

Area of Science:

• Biomedical Engineering
• Cardiovascular Technology
• Signal Processing

Background:

• Hypertension management requires accurate blood pressure (BP) monitoring.
• Photoplethysmography (PPG) is increasingly used in wearable sensors for physiological monitoring.
• Noninvasive BP estimation using PPG signals is an area of significant research interest.

Purpose of the Study:

• To develop and evaluate a novel method for estimating systolic and diastolic BP solely from PPG signals.
• To improve the accuracy of noninvasive BP measurement compared to existing approaches.

Main Methods:

• Feature extraction from PPG signals using the multitaper method (MTM).
• Blood pressure estimation using an artificial neural network (ANN) trained on extracted PPG features.
• Validation of the method's accuracy for systolic and diastolic BP.

Main Results:

• The proposed method demonstrates improved accuracy in BP estimation.
• Mean absolute error for systolic BP was 4.02 ± 2.79 mmHg.
• Mean absolute error for diastolic BP was 2.27 ± 1.82 mmHg.

Conclusions:

• The developed PPG-based method offers a highly accurate approach for noninvasive BP estimation.
• This technique holds potential for integration into wearable devices for continuous hypertension monitoring.
• The combination of MTM and ANN provides a robust framework for PPG signal analysis in BP assessment.