Sleep Temporal Entropy as a Novel Digital Biomarker of Sleep Fragmentation for Cardiometabolic and Mortality Risk

Area of Science:

• Sleep science
• Biomarker discovery
• Cardiometabolic health

Background:

• Sleep fragmentation is an emerging risk factor for cardiometabolic diseases and mortality.
• Current metrics for sleep fragmentation often neglect intra-stage fragmentation and lack clinical validation.
• There is a need for novel biomarkers to accurately quantify sleep fragmentation and its health implications.

Purpose of the Study:

• To develop and validate Sleep Temporal Entropy (STE), a novel biomarker for quantifying sleep fragmentation.
• To assess STE's predictive utility for cardiometabolic outcomes and mortality compared to traditional metrics.

Main Methods:

• Developed STE using Shannon entropy on hypnogram data to measure overall and stage-specific sleep fragmentation.
• Utilized machine learning and Cox regression analyses on two large cohorts (SSHSC and SHHS).
• Evaluated STE's association with diabetes, hypertension, hyperlipidemia, and all-cause/cardiovascular disease (CVD) mortality.

Main Results:

• In the SSHSC cohort, STE demonstrated superior prediction of diabetes, hypertension, and hyperlipidemia compared to existing metrics.
• In the SHHS cohort, STE exhibited U-shaped associations with all-cause and CVD mortality.
• Both lower and higher levels of REM sleep STE were associated with increased mortality risk in the SHHS cohort.

Conclusions:

• Sleep Temporal Entropy (STE) is a novel, non-invasive, and interpretable digital biomarker for sleep fragmentation.
• STE offers improved prediction of cardiometabolic diseases and mortality compared to traditional sleep metrics.
• STE holds promise for clinical applications in assessing and managing sleep-related health risks.