Arterial Blood Pressure Estimation Method from Electrocardiogram Signals using U-Net.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiovascular Physiology

Background:

• Estimating blood pressure from electrocardiogram (ECG) signals is an active research area.
• Existing deep learning models can estimate max, min, and mean arterial blood pressures (ABP) but not continuous ABP.
• There is a need for methods that can estimate continuous ABP from non-invasive ECG signals.

Purpose of the Study:

• To present a novel deep learning method for estimating continuous arterial blood pressure (ABP) from ECG signals.
• To utilize the U-Net architecture for ABP estimation from ECG data.
• To evaluate the accuracy and performance of the proposed ABP estimation method.

Main Methods:

• A deep learning model based on the U-Net architecture was employed.
• The model was trained and evaluated using approximately 185 hours of ECG signal data.
• Performance was assessed by comparing estimated ABP values with ground truth measurements.

Main Results:

• The proposed U-Net model accurately estimated continuous arterial blood pressure (ABP) from ECG signals.
• The accuracy of calculated max, min, and mean ABPs was comparable to previous methods.
• The method demonstrated high accuracy in ABP estimation over extensive signal data.

Conclusions:

• The U-Net deep learning model offers a promising approach for accurate ABP estimation from ECG signals.
• The method provides continuous ABP estimation, surpassing the limitations of previous models.
• Further research is needed to address subject-overfitting and enable practical daily blood pressure monitoring.