A novel computational signal processing framework towards multimodal vital signs extraction using neck-worn wearable...

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Physiological Monitoring

Background:

• Pulse rate (PR) and respiratory rate (RR) are critical vital signs.
• Wearable devices offer convenient monitoring solutions.
• The neck is a potential location for unobtrusive physiological sensing.

Purpose of the Study:

• To develop and validate a methodology for estimating PR and RR using neck-based photoplethysmography (PPG) and accelerometry (Acc).
• To assess the feasibility of using the neck as a multi-modal sensing location for vital signs.
• To improve the accuracy and stability of vital sign estimation from wearable sensors.

Main Methods:

• Utilized a combination of recursive Fast Fourier Transform (FFT)-based dominance scoring and exponentially weighted moving average (EWMA) for signal processing.
• Recorded PPG and Acc signals from healthy participants performing guided breathing and through an altitude generator.
• Calculated rate estimates as bands to incorporate clinical error margins.

Main Results:

• Achieved high accuracy for pulse rate estimation ($$ % within ±5 BPM) compared to gold-standard devices.
• Demonstrated good accuracy for respiratory rate estimation ($$ % within ±3 BrPM) against a visual metronome and ($$ % against a gold-standard device).
• The methodology proved effective even with signals from the neck, an unusual sensing location.

Conclusions:

• The proposed neck-based PPG and Acc methodology provides acceptable PR and RR estimation capabilities.
• This approach offers novel insights for developing medical devices for vital sign monitoring.
• The neck's potential as a multi-modal physiological monitoring site is highlighted.