A Non-Invasive Continuous Blood Pressure Estimation Approach Based on Machine Learning

Area of Science:

• Biomedical Engineering
• Cardiovascular Technology
• Machine Learning in Healthcare

Background:

• Traditional blood pressure (BP) measurement methods face limitations.
• Existing non-invasive continuous BP monitoring techniques require improvement.

Purpose of the Study:

• To establish machine learning models for estimating systolic and diastolic blood pressure.
• To utilize pulse transit time and pulse waveform characteristics for BP estimation.

Main Methods:

• Developed machine learning models for BP estimation.
• Employed the mean impact value method to assess feature importance.
• Utilized a genetic algorithm for parameter optimization.

Main Results:

• The models accurately described the nonlinear relationship between features and BP.
• Achieved high accuracy: 3.27 ± 5.52 mmHg for systolic BP and 1.16 ± 1.97 mmHg for diastolic BP.
• Estimation errors satisfied Advancement of Medical Instrumentation and British Hypertension Society criteria.

Conclusions:

• The proposed machine learning models offer a viable approach for non-invasive continuous BP estimation.
• This research advances the practical application of continuous BP monitoring technologies.