Classification Models for Pulmonary Function using Motion Analysis from Phone Sensors.

Area of Science:

• Biomedical Engineering
• Digital Health
• Pulmonary Medicine

Background:

• Pulmonary function monitoring is crucial for cardiopulmonary patient management.
• Existing methods require specialized equipment and clinical visits.
• Smartphones offer a ubiquitous platform for potential physiological monitoring.

Purpose of the Study:

• To investigate the feasibility of using smartphone sensors to accurately measure pulmonary function.
• To develop and validate improved classification models for pulmonary function assessment.

Main Methods:

• Twenty-four cardiopulmonary patients underwent six-minute walk tests while carrying smartphones.
• Custom software on smartphones recorded motion sensor data during the tests.
• Trained classification models analyzed sensor data to compute pulmonary function metrics.

Main Results:

• The developed model accurately computed pulmonary function for every patient at ten-second intervals.
• The model perfectly classified GOLD stages 1, 2, and 3, aligning with spirometry results.
• This demonstrates the potential for high-fidelity pulmonary function assessment via phone sensors.

Conclusions:

• Smartphone sensor data, analyzed with improved models, can accurately measure pulmonary function.
• This technology supports non-invasive, continuous health monitoring for cardiopulmonary patients.
• Future field trials can explore passive, background monitoring during daily living.