Predicting Pulmonary Function from Phone Sensors.

Area of Science:

• Biomedical Engineering
• Digital Health
• Cardiopulmonary Medicine

Background:

• Pulmonary function is crucial for cardiopulmonary patient health assessment.
• Current monitoring methods often require active participation or clinical settings.
• Smartphones offer ubiquitous potential for passive physiological monitoring.

Purpose of the Study:

• To investigate the accuracy of smartphone motion sensors in predicting pulmonary function.
• To develop and validate machine learning models for pulmonary function classification using phone sensor data.
• To explore the feasibility of continuous, passive pulmonary function monitoring in daily life.

Main Methods:

• Twenty-five cardiopulmonary patients underwent 6-minute walk tests while carrying smartphones.
• Custom software recorded motion data from the phones.
• Spirometry measured each patient's pulmonary function.
• A support vector machine model used signal processing and demographic features to classify pulmonary function levels (e.g., GOLD stages).

Main Results:

• The model achieved high accuracy in predicting pulmonary function categories.
• Classification of Global Initiative for Chronic Obstructive Lung Disease (GOLD) levels 1/2/3 was perfect.
• Distinct motion patterns correlated with different pulmonary function levels.
• Longitudinal changes in pulmonary function were detectable in patients with multiple tests.

Conclusions:

• Smartphone motion sensors can accurately predict pulmonary function.
• This approach supports passive, continuous health monitoring for cardiopulmonary patients.
• The technology shows promise for clinical validation in both hospital and home settings.