Smartphone System for Heart Rate and Breathing Rate Estimation

Abstract

In this short article, we present a new method to use a smartphone placed unattached on a subject’s chest in the supine position to obtain heartbeat and breathing signals and estimate heart and breathing rates, simultaneously. We collected 3-axis accelerometer, gyroscope, and magnetometer signals and performed sensor fusion to extract a user’s breathing signal and breathing rate. A hidden Markov model was used to segment the ballistocardiograph/seismocardiograph signals and extract the heart rate. The smartphone application was verified against breathing belt measurements and electrocardiogram measurements. We modified and proposed several suitable signal quality metrics for seismocardiograph signals. The overall results show that the application accurately estimated the breathing and heart rates, achieving a minimum mean percent error of 2.52% for breathing and 2.33% for heart rate. This work is a big step forward for vital sign estimation using inexpensive pervasive devices.

Keywords

• Ballistocardiography (BCG)
• breathing rate
• heart rate
• machine learning
• seismocardiography (SCG)
• sensor fusion