A new methodology for determining the central pressure waveform from peripheral measurement using Fourier-based...

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Machine Learning in Healthcare

Background:

• Radial applanation tonometry is a key non-invasive method for hemodynamic monitoring.
• Accurate central aortic pressure estimation from radial measurements requires robust mathematical models.
• Wearable healthcare electronics increasingly utilize radial pressure sensing.

Purpose of the Study:

• To develop and validate a novel Fourier-based machine learning (F-ML) methodology.
• To accurately transfer non-invasive radial pressure waveforms to central pressure waveforms.
• To enhance the utility of wearable devices for central hemodynamic assessment.

Main Methods:

• A Fourier-based machine learning (F-ML) transfer function was developed.
• The method was tested using radial and carotid pressure recordings from the Framingham Heart Study (2640 participants).
• Performance was evaluated by correlating F-ML estimates with measured central pressure features and assessing waveform morphology accuracy.

Main Results:

• F-ML estimates showed high correlation with measured systolic (r=0.97), diastolic (r=0.99), and mean (r=0.99) blood pressures.
• Bland-Altman analysis indicated negligible bias, with errors bounded to 5.4 mmHg.
• The F-ML approach achieved an average normalized root mean square error of 5.5% in reconstructing central pressure waveform shape, outperforming conventional methods.

Conclusions:

• The proposed F-ML transfer function provides accurate central pressure waveform estimates from radial measurements.
• This method significantly improves the reconstruction of central pressure waveform morphology compared to existing techniques.
• The F-ML approach holds potential for expanding the use of non-invasive devices in central hemodynamic assessment.