This study is aimed to collect real-time physiological data using two wearable devices (a biometric ring and a biometric wristband), daily lung mechanical measurements by a handheld oscillometer, and participant-reported symptoms in patients with COPD remotely from their home environment. The data will be used to train and validate artificial intelligence and machine learning (AI/ML) models to predict COPD exacerbations in advance of their actual occurrence. The data will also be used to test the new severity classification system for exacerbations of COPD, as well as to determine important relationships between physiological measurements from the wearable devices, the handheld oscillometer, the self-reported symptoms, and the tests performed at the baseline visit.

A single-site, prospective, observational cohort study to collect high-quality multidimensional data from the wearable/portable devices, as well as symptom and exacerbation data, in high-risk patients with frequent exacerbations in order to develop a COPD exacerbation predictive model.

Inclusion Criteria:

• Males/females, age ≥ 40, former/current smokers with ≥10 pack-year smoking history
• FEV1/FVC < 0.7, with 80% < FEV1 ≤50% (moderate, 'GOLD 2') 50% < FEV1 ≤ 30% (severe, 'GOLD 3') or FEV1 < 30% (very severe, 'GOLD 4') COPD
• History of 2 or more exacerbations in the preceding 12 months requiring corticosteroids, antibiotics, or both
• Ability to provide informed consent
• Ability to access internet at least once daily

Exclusion Criteria:

• No existing COPD diagnosis
• Any medical/cognitive/functional condition which renders inability to operate research equipment/devices, and/or to complete daily symptom response

Patients with COPD with a documented history of frequent exacerbations.

While there is no specific method to estimate sample sizes for machine learning-based clinical research, based on two prior COPD studies measuring respiratory rate (RR) differences between stable-state and peak exacerbation phases, it was estimated that detecting a 4 breaths/min difference in RR (effect size 0.74) requires 17 prospectively collected exacerbation events, with alpha set to 0.05, power set to 0.8, and two-tailed analysis. Detecting a more subtle difference (i.e. 2 breaths/min; effect size 0.36) would require 63 events.

Assuming each participant experiences two exacerbations annually, 32 participants would meet these requirements. To account for the long observation period, the potential for non-events, and a 25% attrition rate, up to 50 participants will be recruited for this one-year study.

Due to data privacy regulations, individual participant data collected during this study is not publicly accessible. However, access to anonymized data may be granted upon evaluation by the trial management group. Additional documents will also be available upon inquiry. All requests should be directed to the corresponding author (BAR).