Gyrosphygmogram-Based Blood Pressure Estimation: A Comparative Study of Pulse Transit Time and CNN-LSTM Methods.

Area of Science:

• Biomedical Engineering
• Cardiovascular Technology
• Wearable Health Devices

Background:

• Cuffless blood pressure (BP) monitoring is crucial for continuous cardiovascular health assessment.
• Existing methods often rely on pulse transit time (PTT), which has limitations.
• Novel wearable signals are needed to enhance BP monitoring accuracy.

Purpose of the Study:

• To evaluate the potential of the gyrosphygmogram (GSG) for cuffless BP monitoring.
• To compare GSG-based methods with conventional PTT techniques.
• To develop and assess a deep learning framework for BP estimation using GSG.

Main Methods:

• Simultaneous acquisition of ECG, PPG, and wrist-based GSG signals from 20 healthy adults.
• Calculation of PTT using various signal pairs (ECG-PPG, ECG-GSG, GSG-PPG).
• Development of a multimodal deep learning network fusing temporal and morphological features from all three signals.

Main Results:

• The deep learning model significantly outperformed conventional regression-based PTT methods.
• GSG-ECG fusion within the deep learning model achieved Mean Absolute Errors (MAEs) of 2.89 mmHg (systolic) and 2.2 mmHg (diastolic).
• Performance was validated against three different standards.

Conclusions:

• Integrating GSG into multimodal deep learning substantially enhances cuffless BP estimation accuracy.
• GSG shows significant potential for next-generation wearable devices.
• This approach offers a pathway towards clinically validated accuracy in continuous BP monitoring.