Summary
|
COPD (chronic obstructive pulmonary disease) is a progressive lung disease with an irreversible obstructive air flow and is one of the leading causes of mortality. COPD prevalence and morbidity and their health costs are on the rise, because of the persistence of respiratory symptoms, reduced lung function and frequent exacerbations of COPD (AECOPD) requiring an increased use of healthcare utilization. Moreover, it impairs patients’ quality of life and productivity. Leveraging contact-free and validated portable sensor, a smart watch, and well-characterized cohort of COPD patients, our transdisciplinary team will develop an AI-powered Neural Ordinary Differential Equations (Neural ODEs) and classical local autoregressive models for the early prediction of AECOPD episode(s). Our main hypotheses are that H1) intra- and inter-diurnal pattern of physiological metrics (e.g. heart rate [HR], HR variability [HRV] respiratory rate [RR] and sleep quality, physical activity, saturation O2 [spO2]) change significantly days before an AECOPD event. Thus, identifying these time-lagged changes in physiological metrics can be used to predict individual-specific risk of AECOPD events, H2) changes in exposure to indoor environmental stressors (ES), namely air pollutants, aeroallergens and heat stress, affect changes in physiological measures and H3) ES exposures mediate the effect of physiological metrics on the risk of AECOPD. Identifying early signatures of AECOPD can help develop strategies to engage patients in early interventions to avoid severe episodes of AECOPD. Our findings can be extended for managing other chronic diseases, such as asthma.
|
Team
|
Naresh Kumar (Public Health Sciences), Michael Campos (Pulmonary Medicine), Yan Guo (Public Health Sciences)
|
