Multidimensional sleep profiles via machine learning and risk of dementia and cardiovascular disease.

Abstract

Background: Sleep health comprises several dimensions such as sleep duration and fragmentation, circadian activity, and daytime behavior. Yet, most research has focused on individual sleep characteristics. Studies are needed to identify sleep/circadian profiles incorporating multiple dimensions and to assess their associations with adverse health outcomes.

Methods: This multicenter population-based cohort study identified 24 h actigraphy-based sleep/circadian profiles in 2667 men aged ≥65 years using an unsupervised machine learning approach and investigated their associations with dementia and cardiovascular disease (CVD) incidence over 12 years.

Results: We identify three distinct profiles: active healthy sleepers (AHS; 64.0%), fragmented poor sleepers (FPS; 14.1%), and long and frequent nappers (LFN; 21.9%). Over the follow-up, compared to AHS, FPS exhibit increased risks of dementia and CVD events (HR = 1.35, 95% CI = 1.02-1.78 and HR = 1.32, 95% CI = 1.08-1.60, respectively) after multivariable adjustment, whereas LFN show a marginal association with increased CVD events risk (HR = 1.16, 95% CI = 0.98-1.37) but not with dementia (HR = 1.09, 95%CI = 0.86-1.38).

Conclusions: These results highlight potential targets for sleep interventions and the need for more comprehensive screening of poor sleepers for adverse outcomes.