Multiparametric cloth-based wearable, SimpleSense, estimates blood pressure.

Area of Science:

• Biomedical Engineering
• Cardiovascular Health
• Artificial Intelligence in Medicine

Background:

• Accurate home blood pressure monitoring is crucial for managing hypertension.
• Traditional methods often require calibration and can be cumbersome for daily use.
• Novel approaches using multiparametric vital signs for blood pressure estimation have not been previously reported.

Purpose of the Study:

• To investigate the feasibility of using multiparametric vital signs (ECG, heart sounds, thoracic impedance) combined with machine learning for home blood pressure estimation.
• To compare the accuracy of this novel method against a standard wrist-worn blood pressure cuff.

Main Methods:

• An observational, multi-site study involving 120 participants across different blood pressure categories.
• Simultaneous acquisition of blood pressure measurements (sphygmomanometer) and multiparametric vital signs using the SimpleSense device.
• Training an ensemble of tree-based machine learning models using derived metrics and patient demographics.

Main Results:

• The machine learning model achieved a Mean Absolute Difference (MAD) of ±6.38 mm Hg for systolic blood pressure (SBP) and ±5.10 mm Hg for diastolic blood pressure (DBP).
• These results were comparable, and in some cases superior, to the MAD of a wrist-worn cuff (±8.92 mm Hg for SBP, ±6.86 mm Hg for DBP).
• The models demonstrated the ability to monitor blood pressure without the need for calibration.

Conclusions:

• Multiparametric vital sign data, when analyzed with machine learning, can provide accurate home blood pressure monitoring.
• This technology offers a promising, potentially calibration-free alternative to traditional cuff-based devices for hypertension management.
• Further research can validate this approach for widespread clinical adoption in remote patient monitoring.