A functional supervised learning approach to the study of blood pressure data

Area of Science:

• Biomedical Engineering
• Machine Learning
• Cardiovascular Health

Background:

• Hypertension is a major global health concern.
• Accurate classification of blood pressure is crucial for effective management.
• Traditional methods may not fully capture the dynamic nature of blood pressure over 24 hours.

Purpose of the Study:

• To propose a functional supervised learning scheme for classifying subjects into normotensive and hypertensive groups.
• To utilize solely 24-hour blood pressure data for classification.
• To develop deformable functional models for blood pressure data analysis.

Main Methods:

• Employed functional supervised learning.
• Utilized Fréchet mean and Fréchet variance concepts.
• Developed deformable functional models for blood pressure data.
• Trained the scheme on real clinical data.

Main Results:

• The proposed scheme achieved satisfactory performance in classifying subjects.
• The method effectively uses 24-hour blood pressure data for hypertension classification.
• Deformable functional models proved suitable for blood pressure data.

Conclusions:

• The functional supervised learning scheme provides a viable approach for hypertension classification.
• Sole reliance on 24-hour blood pressure data is effective.
• The method shows promise for clinical application in blood pressure monitoring.