Combining Gaussian Process with Hybrid Optimal Feature Decision in Cuffless Blood Pressure Estimation.

Area of Science:

• Biomedical Engineering
• Cardiovascular Health
• Signal Processing

Background:

• Noninvasive blood pressure estimation is vital for managing cardiovascular diseases and hypertension.
• Continuous blood pressure monitoring is increasingly important, driving interest in cuffless methods.

Purpose of the Study:

• To develop and validate a new methodology for cuffless blood pressure estimation.
• To enhance feature selection for improved accuracy in continuous blood pressure monitoring.

Main Methods:

• A hybrid optimal feature decision (HOFD) framework was proposed to select optimal features.
• Feature selection methods included robust neighbor component analysis (RNCA), minimum redundancy maximum relevance (MRMR), and F-test.
• Gaussian process (GP) was employed as an evaluation criterion to determine the best feature subset.

Main Results:

• The proposed method combining Gaussian process with RNCA achieved lower root mean square errors (RMSEs) for systolic blood pressure (SBP) at 10.75 mmHg and diastolic blood pressure (DBP) at 8.02 mmHg.
• The performance of the proposed algorithm surpassed conventional methods in cuffless blood pressure estimation.
• Experimental results demonstrated the high effectiveness of the developed algorithm.

Conclusions:

• The integration of Gaussian process with hybrid optimal feature decision offers an effective strategy for feature selection in cuffless blood pressure estimation.
• The proposed method provides a promising advancement for accurate and continuous blood pressure monitoring.
• This technique holds potential for improving cardiovascular patient care through reliable noninvasive measurements.