Smartwatch-based ventilatory assessment for COPD screening: A diagnostic accuracy study.

Area of Science:

• Pulmonary Medicine
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Chronic Obstructive Pulmonary Disease (COPD) remains underdiagnosed in China due to limited access to spirometry.
• Smart wearables offer a potential solution for scalable COPD screening through cough and physiological monitoring.

Purpose of the Study:

• To develop and validate a multimodal screening model for COPD using wearable technology.
• To assess the accuracy of the model in differentiating COPD patients from healthy individuals.

Main Methods:

• Machine learning algorithms analyzed cough sounds and smartwatch data (heart rate variability, respiratory rate, oxygen saturation).
• A multimodal model combined cough features and physiological data to predict lung function.
• Model performance was evaluated against spirometry and physician diagnoses.

Main Results:

• The model demonstrated significant correlations between predicted and measured lung function parameters (FVC, FEV1, FEV1/FVC).
• The combined model achieved high accuracy (87.82%), sensitivity (86.96%), and specificity (87.73%) in differentiating COPD from controls.
• Mean absolute errors for FEV1/FVC, FVC%, and FEV1% prediction were 7.4%, 10.6%, and 17.78% respectively.

Conclusions:

• A wearable-based algorithm can effectively screen for ventilatory dysfunction and COPD.
• This technology shows promise for large-scale population screening, potentially reducing the medical burden of COPD.