A recent Mayo Clinic study demonstrates how sleep metrics gathered from wrist-worn activity monitors, when combined with machine learning and standard clinical information, can more accurately forecast whether patients with chronic obstructive pulmonary disease (COPD) will actively participate in a 12-week home-based pulmonary rehabilitation program.
Glimpse:
Published in Mayo Clinic Proceedings: Digital Health, the proof-of-concept research led by Dr. Stephanie Zawada showed that adding a “Composite Sleep Health Score” derived from one week of pre-rehabilitation wearable data significantly boosted the predictive performance of machine learning models. This approach helps identify COPD patients who may need extra support to complete remote rehab, addressing the common problem of low adherence caused by sleep disturbances and fatigue.
Researchers at Mayo Clinic have introduced an innovative method to tackle one of the biggest hurdles in managing chronic obstructive pulmonary disease: low patient participation in pulmonary rehabilitation. Many individuals with COPD struggle with breathing difficulties that also disrupt their sleep, leaving them tired and less likely to stick with structured exercise and education sessions especially in fully remote, home-based programs that typically last 12 weeks.
In this study, scientists asked participants to wear wrist-based activity monitors for one week before beginning their home pulmonary rehabilitation. The devices captured detailed sleep patterns, which the team used to calculate a Composite Sleep Health Score reflecting overall sleep quality, duration, and efficiency. This score was then integrated with conventional clinical factors such as age, sex, smoking status, comorbidity index, breathing test results, and symptom severity scores. Machine learning algorithms analyzed the combined dataset to predict how consistently patients would engage throughout the full three-month program.
The findings revealed that incorporating the wearable-derived sleep information noticeably enhanced prediction accuracy in several models, including support vector machines and logistic regression. For instance, the inclusion of sleep metrics improved specificity and overall accuracy compared to models relying solely on traditional clinical data. This proof-of-concept work suggests clinicians could soon use simple wearable tracking to spot patients at higher risk of dropping out early and provide them with tailored coaching or additional resources right from the start.
Conducted with institutional review board approval, the research builds on Mayo Clinic’s ongoing efforts to make pulmonary rehabilitation more accessible and effective through digital tools and health coaching. Experts believe this wearable-plus-machine-learning strategy could play a key role in personalizing remote care for COPD, ultimately helping more patients improve their breathing, energy levels, and quality of life while reducing hospital readmissions.
“Our goal was to explore how wearable data could improve dropout rates in remote pulmonary rehabilitation programs.”
By
HB Team
Related News
