Body position classification for cardiorespiratory measurement.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Physiological Monitoring

Background:

• Heart rate variability and respiratory function are influenced by body posture.
• Impedance pneumography for respiratory monitoring requires position-dependent calibration.
• Smartphone sensors offer a potential solution for unobtrusive physiological monitoring.

Purpose of the Study:

• To assess the accuracy of smartphone sensors in determining body position.
• To develop algorithms for cardiorespiratory monitoring independent of body position.
• To evaluate the feasibility of using smartphone data for long-term physiological studies.

Main Methods:

• Collected data from 24 students in various positions (lying, sitting, standing).
• Utilized iPhone's accelerometer, gyroscope, and magnetometer to calculate body attitude.
• Trained and cross-validated five algorithms using sensor data for position classification.

Main Results:

• Achieved 100% accuracy in differentiating major positions (lying vs. sitting/standing) with all algorithms.
• Random forest classifier reached 90% accuracy differentiating sitting and standing.
• A hybrid classifier achieved 99% accuracy for online measurement without segmentation.

Conclusions:

• Smartphone sensors can accurately determine body position, crucial for physiological monitoring.
• Developed algorithms show promise for position-independent cardiorespiratory monitoring.
• This approach facilitates unobtrusive, long-term health studies using readily available technology.