Remote COPD Severity and Exacerbation Detection Using Heart Rate and Activity Data Measured from a Wearable Device

Area of Science:

• Pulmonary Medicine and Digital Health
• Biomedical Engineering and Wearable Technology

Background:

• Chronic obstructive pulmonary disease (COPD) is a major global health concern, necessitating improved monitoring strategies.
• Traditional COPD severity assessment relies on clinical settings, limiting real-time insights into disease progression and autonomic dysfunction.
• Remote monitoring via wearable devices offers a novel approach for unobtrusive, continuous patient data collection in daily life.

Purpose of the Study:

• To investigate the efficacy of low-cost wearable devices in predicting COPD severity and exacerbations.
• To evaluate the potential of using heart rate and activity data from smartwatches for COPD management.

Main Methods:

• Collected smartwatch sensor data (heart rate, activity) from 35 COPD patients over three months.
• Utilized early, continuous, unobtrusive wearable data for disease trajectory prediction.
• Employed an Isolation Forest model for exacerbation prediction based on extracted wearable features.

Main Results:

• Future COPD disease trajectory can be predicted using initial days of wearable data.
• The Isolation Forest model achieved an Area Under the Curve (AUC) of 0.69 for exacerbation prediction.
• This performance demonstrates an improvement over random chance classification.

Conclusions:

• Continuous, unobtrusive monitoring with low-cost wearable devices can effectively predict COPD severity and exacerbations.
• Early detection of disease trajectory using wearable data holds significant potential for proactive COPD management.
• This technology offers a promising advancement in remote patient monitoring for chronic respiratory diseases.