A novel deep learning based automatic auscultatory method to measure blood pressure

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiovascular Diagnostics

Background:

• Accurate automatic blood pressure (BP) measurement is crucial for clinical practice.
• Existing automated methods have limitations, necessitating innovative techniques.

Purpose of the Study:

• To present and evaluate a novel deep learning-based automatic BP measurement method.
• To assess the impact of stethoscope position and contact pressure on BP measurements.

Main Methods:

• A convolutional neural network (CNN) was developed to detect Korotkoff sounds beat-by-beat.
• A mapping algorithm correlated identified sounds with cuff pressures for systolic BP (SBP) and diastolic BP (DBP) calculation.
• The method's accuracy was validated against manual auscultation, analyzing variations due to stethoscope placement and pressure.

Main Results:

• The proposed method achieved low overall measurement errors: 1.4 ± 2.4 mmHg for SBP and 3.3 ± 2.9 mmHg for DBP.
• Minimal SBP differences were observed across different stethoscope positions and contact pressures.
• DBP measurements were significantly lower when the stethoscope was placed under the cuff (2.0 mmHg difference, P < 0.01).

Conclusions:

• The deep learning-based method is an effective technique for automatic BP measurement.
• This novel approach has the potential to advance beyond current oscillometric methods for BP monitoring.